New Business Combinations Accounting Rules and the Mergers and Acquisitions Activity

NEW BUSINESS COMBINATIONS ACCOUNTING RULES AND THE MERGERS AND ACQUISITIONS ACTIVITY

HUMBERTO NUNO RITO RIBEIRO

Doctor of Philosophy

DE MONTFORT UNIVERSITY LEICESTER BUSINESS SCHOOL

November 2009

This copy of the thesis has been supplied on condition that anyone who consults it is understood to recognise that its copyright rests with its author and that no quotation from the thesis and no information derived from it may be published without proper acknowledgement. Leicester Business School, De Montfort University

New Business Combinations Accounting Rules and the Mergers and Acquisitions Activity

Humberto Nuno Rito Ribeiro

Doctor of Philosophy

2009

Thesis Summary

The perennial controversy in business combinations accounting and its dialectic with stakeholders’ interests under the complexity of the Mergers and Acquisitions (M&A) activity is the centrepiece of analysis in this thesis. It is argued here that the accounting regulation should be as neutral as possible for the economic activity, although it is recognised that accounting changes may result in economic effects. In the case of the changes for business combinations accounting in the USA, lobbying was so fierce that in order to achieve the abolition of accounting choice in M&A accounting, it forced the standard-setter to compromise and to change substantially some of its earlier proposals. Such fierce lobbying cast doubts about whether it was effectively possible to mitigate such economic effects, resulting in a possible impact of the accounting changes on the M&A activity.

The occurrence of M&A in waves is yet to be fully theorised. Nevertheless, existing literature established relationships between M&A activity and some key economic and financial factors, and has provided several interesting theories and other meaningful contributions for this thesis. It was therefore possible to examine whether the changes in the accounting rules produced any significant impact on the M&A activity.

The findings obtained from the testing of the research hypotheses suggest that the new M&A accounting rules did not result in significant impacts on overall M&A activity. Nevertheless, from the study of managers’ perceptions, and from the examination of annual reports of S&P 500 companies, a considerable impact on the financial reporting was found.

Key words: Mergers and Acquisitions (M&A), M&A activity and waves, Accounting regulation, Economic consequences, Business combinations, Accounting choice, Pooling of interests method, Purchase method, Goodwill amortisation, Goodwill impairment

Dedication

This thesis is dedicated to my family in Portugal and in Brazil, in thanks for their continuing support and encouragement during its preparation. Iria, Franklin, Stella, Nita, and godmother Lúcia, I love you all.

In memory of José Maria Pereira

who could not live long enough to witness the accomplishement of this thesis. Many thanks for everything, may you enjoy eternal life.

Post scriptum: I had the confirmation of my PhD on my Godfather’s first anniversary

Acknowledgements

I would like to acknowledge in particular my director of studies, Professor David Crowther, Leicester Business School, De Montfort University, for his supervision and constant support. I also would like to acknowledge Peter Scott, De Montfort University, and Bode Akinwande, London Metropolitan University, for their advice during the development of the research; Maria Simatova, London Business School, for her assistance with data collecting and treatment; and Ken Westmoreland, London, and Rachel Alves, Chichester, for their continuous reviews; and, finally, Stuart Cooper, Aston University, and Ashok Patel, De Montfort University, for their substantive reviews. Finally, yet not least important I address my recognition to other friends and colleagues, anonymous referees, discussants and conference participants, which helped in some way to develop this major research undertaking.

I also would like to acknowledge the funding for this research and for presentations in seminars and conferences obtained from the following programs and institutions: PRODEP, Ministério da Ciência e do Ensino Superior, Portugal and EU ESTiG, Instituto Politécnico de Bragança, Portugal Leicester Business School, De Montfort University, UK Fundação para a Ciência e a Tecnologia, MCTES, Portugal Fundação Calouste Gulbenkian, Portugal London Metropolitan University, UK Universidad del País Vasco, Spain The Institute of Chartered Accountants in England and Wales, UK

Finally, I acknowledge the institutions that provided training and logistical support during the early stage of the research: Universidad de Santiago de Compostela, Spain Universidad Carlos III, Madrid, Spain

Contents

Page

Chapter 1 Introduction 16

1.1 Summary 16

1.2 The development of accounting theory and regulation 17

1.3 Ontology, theoretical identification and theory validation 31

1.4 Accounting, finance, and globalisation 35

1.5 Conclusions 38

Chapter 2 Political Nature of Accounting Standard Setting and Developments on Business Combinations Accounting 43

2.1 Introduction 43

2.2 Towards a regulated conceptual framework for accounting 44

2.3 Lobbying and political influences on standard setting 55

2.4 Business combinations in the USA: an under pressure accounting issue 67

2.5 Accounting choice and business combinations accounting 80

2.6 Lobbying and pressures on FASB’s new M&A accounting proposals 86

2.7 Conclusions 113

Chapter 3 M&A Activity and M&A Accounting 115

3.1 Introduction 115

3.2 Terminology and definitions 116

3.3 M&A activity pattern and M&A waves 118

3.4 New age for business combinations and goodwill accounting 129

3.5 Neutrality versus economic effects in M&A accounting 137

3.6 Possible impacts on M&A activity and M&A accounting 143

3.7 Conclusions 146

Chapter 4 Hypotheses 148

4.1 Introduction 148

4.2 M&A activity during the 2000-2002 period 150

4.3 M&A activity during the 1994-2008 period 157

4.4 Research methodology 159

4.5 Factors and theories explaining M&A activity 163

4.6 Reasons explaining M&A abandonment 167

4.7 Conclusions 171

Chapter 5 Survey and Financial Reporting Analysis 173

5.1 Introduction 173

5.2 Survey on companies’ reactions to the new M&A accounting rules 174

5.3 Analysis of annual reports 189

5.3.1 Methodology of analysis 190

5.3.2 Data sources 190

5.3.3 Data collection 192

5.3.4 SFAS 142 impacts sample 208

5.3.5 Basic descriptive statistics and analysis 212

5.3.6 Cross-sectional analysis 215

5.4 Conclusions 220

Chapter 6 Data Collection 223

6.1 Introduction 223

6.2 Data sources 224

6.3 M&A data selection 227

6.4 M&A sample 234

6.5 Data aggregation 236

6.6 Conclusions 243

Chapter 7 Models Development and Testing Results 244

7.1 Summary 244

7.2 Introduction 245

7.3 Construction of metrics 252

7.4 Statistical models for hypotheses testing 259

7.5 Variable definitions and predictions 264

7.6 Univariate descriptive statistics 276

7.7 Empirical tests and discussion of results 280

7.7.1 Bivariate analysis 280

7.7.2 Multivariate analysis 284

7.7.3 Outliers and influential points 303

7.7.4 Sensitivity analysis 312

7.7.5 Forecasting model validation 319

7.8 Conclusions 321

Chapter 8 Discussion and Interpretation 323

8.1 Introduction 323

8.2 Accounting regulation and accounting choice: an international perspective for business combinations 325

8.3 Accounting choice in business combinations accounting and M&A activity 336

8.4 New M&A accounting rules and M&A activity: business as usual? 342

8.5 Conclusions 344

Chapter 9 Conclusions and Further Research 346

9.1 Introduction 346

9.2 Limitations 347

9.3 Summary of main research and generalisation 350

9.4 Suggestions for further research 353

9.5 Contributions 354

References 359

Appendix A S&P 500 Companies List as of 31 December 2004 417

Appendix B Questionnaire Addressed to S&P 500 companies 430

Appendix C Crosstabulations for Questionnaire Data 431

Appendix D Goodwill and Other Intangible Assets (OIA), and Impact on Diluted EPS, by Industry 437

Appendix E Descriptive Statistics 438

Appendix F Pearson/Spearman Correlation Matrixes 442

Appendix G Regression Analysis for Hypotheses One and Two 446

Appendix H Residuals’ Autocorrelation Tables and Correlograms 453

Appendix I Durbin-Watson Distribution and Critical Values 467

Appendix J Plots for Normal Distribution of Residuals 469

Appendix K Outliers and Influential Points 475

Appendix L Sensitivity Analyses 487

Tables

Page

Table 3.1 Pro forma impact on EPS in 2001 of selected S&P 500 companies 145

Table 5.1 Estimated SFAS 142 impacts on diluted EPS by industry 216

Table 6.1 Summary of some major M&A data sources for the USA 225

Table 6.2 Sample description for the 2000-2002’s period 235

Table 6.3 Sample description for the 1994-2008’s period 236

Table 7.1 First lag autocorrelations and Durbin-Watson statistic values 294

Table 7.2 Key tests for residuals’ normality 299

Table 7.3 Tests for heteroscedasticity 303

Table 7.4 Outliers diagnostics 308

Table 7.5 Resume of samples for sensitivity analysis 315

Table 7.6 Regressions’ sensitivity to abnormal and non-trading day’s removals 316

Table 7.7 Sensitivity analysis using alternative event windows 317

Table 7.8 Out-of-sample accuracy measurement 321

Table 8.1 Accounting for business combinations worldwide in the 1990’s 327

Table 8.2 Accounting for M&A in Europe and in the USA: 1999-2000 332

Table C.1 Crosstabulation for SFAS 141 and IT & Financials industries 431

Table C.2 Crosstabulation for SFAS 142 and IT & Financials industries 432

Table C.3 Crosstabulation for SFAS 141 and IT industry 433

Table C.4 Crosstabulation for SFAS 142 and IT industry 434

Table C.5 Crosstabulation for SFAS 141 and Financials industry 435

Table C.6 Crosstabulation for SFAS 142 and Financials industry 436

Table E.1 Descriptive statistics for unadjusted and non-dummy variables 438

Table F.1 Pearson/Spearman correlation matrixes for coefficient estimates 442

Table G.1 Regression models outputs and tests results for hypothesis one 446

Table G.2 Regression models outputs and tests results for hypothesis two 450

Table G.3 Résumé of regression analysis for hypothesis one and two 452

Table H.1 Autocorrelations for hypothesis one 453

Table H.2 Autocorrelations for hypothesis two 457

Table H.3 Partial autocorrelations for hypothesis one 459

Table H.4 Partial autocorrelations for hypothesis two 463

Table I.1 Critical Values for the Durbin-Watson test 468

Table K.1 Unusual residuals for hypothesis one 475

Table K.2 Unusual residuals for hypothesis two 477

Table K.3 Influential points for hypothesis one 483

Table K.4 Influencial points for hypothesis two 486

Table L.1 Additional regression model estimation and tests results 488

Table L.2 Estimated autocorrelations for additional regression’ residuals 489

Table L.3 Estimated partial autocorrelations for additional regression’ residuals 490

Figures

Page

Fig. 2.1 Goodwill recognised in business combinations, 1967-2001 77

Fig. 2.2 Average goodwill amortisation period in the USA, 1985-2001 89

Fig. 3.1 Number of mergers completed in the USA, 1895-1990 119

Fig. 3.2 Monthly M&A completed deals and dollar values, 1981-2005 123

Fig. 3.3 Pooling versus purchase in the USA, 1967-2001 130

Fig. 3.4 Percentage of deals using pooling of interests versus purchase method, in dollar value, 1998-2000 133

Fig. 3.5 Monthly M&A completed deals in number of deals and dollar values in the USA, from 01-2000 to 05-2005 143

Fig. 4.1 Number of announced M&A deals, in the period 2000-2002, during weekdays 151

Fig. 4.2 Number of announced M&A deals in the thirty days surrounding the event date 152

Fig. 4.3 Number of M&A announcements during weekends in 2001 154

Fig. 4.4 Quarterly number of M&A deals announced and completed during the 1994-2008 period 158

Fig. 5.1 Information on the respondents’ corporate sector 179

Fig. 5.2 Information on the respondents’ professional background 180

Fig. 5.3 Impact of the new accounting standards on the completion of M&A deals 181

Fig. 5.4 Relevance and impact of the new accounting standards on the M&A decision-making and overall activity 186

Fig. 5.5 S&P 500 index companies by industry as of 31 December 2004 209

Fig. 5.6 Annual report sample companies by industry 212

Fig. 6.1 Yearly number and value of M&A deals in the USA, 1990-2002 226

Fig. D.1 Weighted average goodwill and OIA, and diluted EPS, for 2000-01 437

Fig. D.2 Total amounts of goodwill and OIA added back for 2000-01 437

Fig. H.1 Autocorrelations correlograms for hypothesis one 465

Fig. H.2 Autocorrelations correlogram for hypothesis two 466

Fig. I.1 The five regions of the Durbin-Watson d-statistic 467

Fig. J.1 Plot of residuals for hypothesis one 469

Fig. J.2 Plot of residuals for hypothesis two 470

Fig. J.3 Histograms of residuals with normal distribution curve superimposed for hypothesis one 471

Fig. J.4 Histogram of residuals with normal distribution curve superimposed for hypothesis two 472

Fig. J.5 Residuals’ normal probability plots for hypothesis one 473

Fig. J.6 Residuals’ normal probability plot for hypothesis two 474

Fig. K.1 Box-and-Whisker plot of residuals for hypothesis one 478

Fig. K.2 Box-and-Whisker plot of residuals for hypothesis two 479

Fig. K.3 Plot of outliers for hypothesis one 480

Fig. K.4 Plot of outliers for hypothesis two 482

Fig. L.1 Box-and-Whisker plot of residuals with major outliers removed 487

Fig. L.2 Outlier plot with major outliers’ elimination 487

Fig. L.3 Autocorrelations correlogram for additional regression model 490

Chapter 1 Introduction

1.1 Summary

This thesis is concerned with the possible economic consequences that may arise from changes in accounting regulation. More precisely, it investigates whether the changes occurred in business combinations accounting in 2001 affected the Mergers & Acquisitions (M&A) activity in the USA.1

Accounting can be perceived from different perspectives as it serves different users and purposes. For example, Hendriksen & van Breda (1992) refer to the following approaches for accounting: tax, legal, ethical, economic, behavioural, and structural. Different approaches and different influences which can be ultimately tracked from four thousand years of previous accounting forms, being the major milestone the development of bookkeeping during the Italian renaissance, which the Franciscan friar Luca Pacioli would formulate and publish in Venice in 1494.2

1 Vid. the note on M&A and business combinations definitions and terminology in chapter 3. 2 Luca Pacioli is also known as Luca Paciolo, Lucca Paccioli, Luca Paciuolo, Luca di Borgo, among other names. He compiled the existing knowledge about double-entry bookkeeping in the Tractatus Particularis de Computis et Scripturis, which formed part of the Summa de Arithmética, Geometria, Proportioni e Proportionalita, a broader scientific work which included several topics, mostly related with mathematics.

For the purposes of the present research, accounting is important from a financial point of view, as it is concerned with financial reporting. One could therefore argue that more important to discuss in depth some specific theoretical aspects of accounting would be to focus exclusively on the process and the consequences of the accounting regulatory process, particularly for all the parties with interests related to corporate reporting. However, although it may resemble a paradox, together with the study of the developments in the accounting regulation, such desiderate also implies a reference to the recent evolution in the accounting theory, as both matters are intrinsically linked.3

On the other hand, M&A activity, which like accounting is also interdisciplinary, is closely related with finance and economics, and constitutes a complex social sciences topic. Therefore, a brief epistemological review on social sciences follows, along with an historical review in accounting and exchange markets, which will help to understand contemporary accounting and finance theories, and their interlinked relationships. This examination provides a starting point for the present research, as it deals with the idiosyncrasies of M&A accounting, such as its politics and lobbying. The political and lobbying influences effects on accounting regulation is a topic only briefly introduced in this chapter, as it is to be examined in depth in chapter 2, together with its effects on business combinations accounting. It also launches the basis for the discussion as to why it was not possible until now to construct a comprehensive conceptual theory for accounting.

1.2 The development of accounting theory and regulation

From Pacioli’s time, double-entry bookkeeping remained barely unchanged as the basic technique in accounting (Hendriksen & van Breda, 1992; Kam, 1990). However, the world suffered major revolutions since then, first with the

3 For the development of this topic we rely primarily on Hendriksen & van Breda (1992), with no disregard to other major works on accounting theory, such as the ones of Kam (1990), or Belkaoui (1985).

17 maritime discoveries, and later with the industrial surge.4 Such discoveries would result in an immense flow of maritime trade, and also in an increasing demand for funds to start new commercial routes and to launch other discoveries’ enterprises. This motivated individuals to associate in ventures, which would lead to the creation of joint stock companies. Monetarism developed alongside maritime trade and the emerging trading companies. Later, the emergence of exchange markets made it possible to trade the stock of joint companies.5 However, the exchange markets were still incipient and somewhat naïve, as the shortfall of information and the relatively small amounts of capital involved made it easily vulnerable to manipulation. Therefore, massive bubbles and colossal bursts occurred, resulting in some remarkable losses even for many notable individuals.6

It was only in the 19th century that accounting started to be subject to some regulation (Hendriksen & van Breda, 1992). The first step was made in the UK, with the publication of the Joint Stock Companies Act in 1844, which called for auditors to ensure completeness and fairness of the balance sheets disclosed by companies. It also required companies to maintain accounts. By improving the quality of the information disclosed, as the auditors could also hire accountants and other experts, a better functioning from the exchange markets was expected, since investors could be better informed about the companies.7

4 Six years before Paccioli’s tractatus, Bartolomeu Dias crossed the Cape of Good Hope discovering the maritime way from Europe to Asia, and enabling the maritime route to India, which Vasco da Gama would sail ten years later. Only four years after Good Hope’s breakthrough, Colombus would discover America. 5 The first stock exchange was founded in Amsterdam, Netherlands, in 1602. In the UK, the London Stock Exchange was created in 1773, and in the USA, the Philadelphia exchange was the first to be created, in 1746, followed by the New York Stock Exchange (NYSE), in 1792. 6 When the South Sea Company bubble burst in the 18th century, even the British Royal family lost a fortune of a considerable amount. For that reason, stock certificates and joint stock companies were banned from the UK, for a period of over a century, until the cessation of the so-called Bubble Act in 1825 (Hendriksen & van Breda, 1992: 46; 63). 7 For a review of accounting in the UK, in the form of an “archaeology of financial reporting”, vid. Crowther (2002b).

In the USA, the corporate and financial development from the post-World War I period resulted in an increased pressure from the financial sector and from the stock exchanges, which led to a shift in the objectives of the existing basic accounting structure. The securities markets crash in 1929 would reinforce such pressure, which, alongside with important issues regarding financial reporting, led to a decrease of importance and participation in the accounting establishment from the accounting profession itself. From a previous accounting concerned primarily with the interests of creditors and management, it was now demanded a shift to a new group of stakeholders: investors and shareholders. According to Hendriksen & van Breda (1992: 98):

“The change in the objective of financial statements led to: 1. A de-emphasis of the balance sheet as a statement of values. 2. A consequent increased emphasis on the income statement and a uniform concept of income. 3. A need for full disclosure of relevant financial information, by presenting more complete financial statements and increasing the use of footnotes. 4. An emphasis on consistency in reporting, particularly with respect to the income statement.”.8

Consequently, corporate development and economic events led to a quest for accounting principles that could be coherently integrated in a conceptual framework. This would become a matter of utter importance, but from the start, the intrinsic complexity of accounting seemed to make this purpose unlikely to concretize.

In the scope of the accounting richness and diversity, it would never be easy to settle a comprehensive and unanimous theoretical framework (vid. Hendriksen & van Breda, 1992). As argued by many authors, an unanimous framework is an impossibility as consensus can never be reached (see e.g. Seidler, 1984;

8 In an interesting analogy, Ripley (1927) linked the balance sheet to a still photograph of the situation of a company at a certain point in time. This analogy would be later refuted by May (1934), arguing that is not possible to take a picture of history. Additionally, he criticised the accuracy of the accounting records that were produced by then.

Gerboth, 1987). Even if consensus were possible, it would be undesirable, as it would never be possible to consider comprehensively the whole complexity inherent to accounting and to its heterogeneous and dynamic environment, which is typical for a discipline that belongs to the group of the social sciences, in a single framework.9

The way in which the accounting is constructed does not help either (vid. e.g. Hendriksen & van Breda, 1992). One could argue that accounting can be primarily developed in two ways: bottom-up and top-down. Bottom-up as the accounting practice was in early times the source of the generally accepted accounting principles (GAAP).10 However, new challenges for the accounting profession were raised by the economic and entrepreneurial development of the early 20th century, a natural outcome following the Industrial Revolution and the advent of capitalism.11 Additionally, the impressive events of the Great Depression, which resulted in an extraordinary number of corporate bankruptcies, many of which including accounting frauds, shed light on the discrepancies of the existing accounting practice. In the absence of detailed codes of procedure and well-established and coherent accounting principles, this led to a condition where the financial reporting was substantially relying on the professional judgement. Moreover, adding a subtle rationale, one could even suggest that the existing condition of accounting and financial reporting at that time could be easily manipulated according to the interests of owners and managers, possibly misleading external entities, such as investors and creditors. In conclusion, in face of anecdotal evidence of a deficit of uniformity in financial reporting, linked to biased accounting practices, it was hazardous for the users

9 For examinations regarding the complex nature of accounting vid. e.g. Mattessich (1995), or Hendriksen & van Breda (1992). 10 Due to the complexity of accounting theory, which has always resulted in a shortfall of consensus on accounting principles, the GAAP definition resulted to be somewhat vague, and therefore difficult to define. Roughly, one can understand GAAP as accounting practices accepted by the accounting community and with substantial acceptance from the accounting regulators. A formal definition of this concept is shown later in this chapter. 11 Vid. Hendriksen & van Breda (1992) for a review of the development of accounting thought. This and the following three paragraphs rely mostly on Hendriksen & van Breda’s work.

20 to rely on the accounting information. The accounting profession was therefore under pressure, and a major shift was required.

Modern financial accounting in the USA is marked by the 1929 Stock Exchange Crash and the subsequent Great Depression. The turmoil in the economy and in the financial markets led the US Congress to approve the Securities Act of 1933, and the Securities Exchange Act of 1934, which brought to life the Securities and Exchange Commission (SEC). The SEC was created to supervise and to regulate the securities and exchange markets. Additionally, it has also been given regulatory authority to the SEC in order to set accounting rules. However, the SEC would delegate this responsibility to private accounting committees and boards. From 1936 to 1973, the accounting policy was delegated to the American Institute of Certified Public Accountants (AICPA), and to its predecessor, the American Institute of Accountants (AIA).12 However, as AICPA’s decisions and hesitations started to face increased criticism, it would be replaced in such role in 1973 by the Financial Accounting Standards Board (FASB).13 After FASB induction, the accounting construction process become

12 The first professional body of accountants formed in the UK was the Society of Accountants in Edinburgh, Scotland, in 1853 (Kam, 1990). A few years later the designation of “chartered accountant” would be adopted by three Scottish societies (Kam, 1990: 28). Other associations would follow across the UK and, in 1880, five of the existing organisations would be incorporated in the newly founded Institute of Chartered Accountants in England and Wales, in an attempt to remove “an increasing number of unqualified people doing accounting-auditing work” (Kam, 1990: 28). The first accounting professional organisation to be formed in North America was in Montreal, Canada, in 1880, followed by the Institute of Accountants and Bookkeepers in the USA in 1882 (Kam, 1990). 13 More precisely, the accounting regulatory policy was in charge of the AICPA from 1887 to 1934; the SEC from 1934 to 1936, the Committee on Accounting Procedure (CAP) from 1936 to 1959, and the Accounting Principles Board (APB) from 1959 to 1973. The AICPA was created as the American Association of Public Accountants (AAPA) in 1887, and it was reorganized later in 1917 as The American Institute of Accountants (AIA), before adopting its current designation in 1957, when “Certified Public” was added to its name (Hendriksen & van Breda, 1992). Both CAP and APB were created and controlled by the AICPA. While AICPA is a private institution more accounting professionals-orientated, the FASB is also a private organization, but independent from the AICPA, and primarily user-orientated.

21 essentially top-down: from standard setting to the accounting profession. Nevertheless, it was possible to maintain accounting experts directly involved in the regulation, and the accounting professionals continued to be invited to participate in the standard-setting process.

Regardless the status of the accounting profession and regulation, the development of accounting theory remained an issue, facing more drawbacks than advances. One of the first academics in the USA interested in the development of a broad set of accounting principles was Professor William Paton, founder and president of the American Accounting Association (AAA), who published his doctoral thesis entitled “Accounting theory” in 1922.14 Afterwards, when he was the research director of the AAA, he published a first essay of a statement in The Accounting Review (Paton, 1936). This would be the first of a series of monographs focused on accounting principles, which would also have the later contribution of Professor Ananias Littleton (Paton & Littleton, 1940). However, such efforts have never resulted in any definitive accounting statement or standard, but only in tentative statements, as it was not possible to reach a consensus amongst the academic and professional accounting community.

The post-World War II boom also seemed to have boosted the quest for accounting principles, and not only by the AAA, but also from the accounting policy setter, which was by then in charge of the AIA’s (AICPA) CAP.15 Later, when the AICPA’s Accounting Principles Board (APB) was formed, a permanent study group on the accounting principles task was formed. As Professor Maurice Moonitz, Director of Accounting Research of the AICPA, was

14 Among the most prominent early contributors for the accounting theory, it is possible to find names such as Cole (1908), Sprague (1908), and Hatfield (1909). Zeff (2000: 108) refers to them as the authors of the “three most important book-length works on accounting theory and practice during the first decade of the century in the United States”. Other later relevant contributions included similar approaches, such as the one of Montgomery (1912), and broader ones, such as the economic one of Canning (1929). 15 This boost was also felt in business combinations accounting as to be discussed later in chapter 2.

22 commissioned to research on the accounting principles and postulates, it would then produce the Accounting Research Study (ARS) No. 1, The Basic Postulates of Accounting, published in 1961. Another reference work, the ARS 3, A Tentative Set of Broad Accounting Principles for Business Enterprises, authored by Moonitz and Robert Sprouse, a future FASB member, was published one year later, in 1962.16 These two studies were focused on the basic accounting postulates, and on the accounting principles, respectively. ARS 1 considered postulates as basic assumptions concerning the accounting environment for which the accounting would need to adjust in order to operate. This position was substantially different from the one expressed previously by the AIA’s Committee on Terminology, in the Accounting Terminology Bulletin (ATB) No. 1, Review and Resume, published in 1953. Among the different generic definitions of principles that could be found in an English dictionary, the committee found that “A general law or rule adopted or professed as a guide to action; a settled ground or basis of conduct or practice” was the one shared by most of the accountants, particularly by the “practicing public accountants”.17 As for postulates, the committee claimed that (paragraph 17, ATB 1):

“Initially, accounting postulates are derived from experience and reason; after postulates so derived have proved useful, they become accepted as principles of accounting. When this acceptance is sufficiently widespread, they become a part of the "generally accepted accounting principles" which constitute for accountants the canons of their art. It is not convenient, either in conversation or in writing on accounting subjects, to add "(meaning number three)" each time the word principle is used, though that essentially is understood.”.18

It was argued that if principles of accounting were just rules, it should be possible then to deduce them from the more basic assumptions called postulates.19

16 Robert Sprouse would become a member of the newly created FASB in 1973. 17 Paragraphs 16-17, ATB 1. 18 Also of interest, the authors views on accounting as an “art”. 19 Chambers (1963) op. cit. Hendriksen & van Breda (1992: 101).

Not surprisingly, the reactions from the accounting community to both ARS 1 and ARS 3 were not favourable, and they failed to be accepted. Critics from individual members of the APB were placed in the end of both ARS’s documents. For example, Leonard Spacek rejected the idea of several postulates, arguing in favour of the existence of a single one (Moonitz, 1961: 57).20 The APB published its position in the Accounting Principles Board Statement (APS) No. 1, Statement by the Accounting Principles Board, in 1962b, stating that (paragraph 2, APS 1):

“Prior to its publication, Study No. 3 has been read and commented upon by a limited number of people in the field of accounting. Their reactions range from endorsement of the ideas set forth in the study of "Broad Principles" to misgivings that compliance with the recommendations set forth by the authors would lead to misleading financial statements. The Board is therefore treating these two studies (the one on "Postulates" and the other on "Principles") as conscientious attempts by the accounting research staff to resolve major accounting issues which, however, contain inferences and recommendations in part of a speculative and tentative nature.”.

Accordingly, the APB claimed that (paragraph 3, APS 1):

“while these studies are a valuable contribution to accounting thinking, they are too radically different from present generally accepted accounting principles for acceptance at this time.”.

Afterwards, Deinzer (1965) also criticized the lack of a link between postulates and principles. Additionally, in a later ARS 7, Grady (1965), employing a pragmatic-inductive method instead of the deductive method of ARS 1 and ARS 3, challenged the idea of a single and uniform accounting model, recalling the complexity and diversity in accounting.21

20 Leonard Spacek was member of the APB and managing partner of Arthur Andersen. 21 Vid. the epistemological piece shown later in this chapter.

Following widespread criticism from virtually all quadrants, ARS 1 and ARS 3 were rejected, including by the AICPA’s APB. Although welcomed by the accounting community, ARS 7 would not have a much better fate, as none of these series of studies would become a statement of accounting principles. It became evident that the quest for broad accounting principles returned once again to ground zero. Nonetheless, the criticism provoked by these studies brought some new interesting ideas to the foreground.

In 1963, Vatter argued that instead of principles or postulates, the focus should be primarily on objectives. This was an early indication of a major shift, which would materialise in a reference document published by the AAA in 1966: A Statement of Basic Accounting Theory, which would became best known by its acronym, ASOBAT. Additionally, in ASOBAT it was argued that accounting should allow its users to make informed judgements in order to make decisions. This user-orientated view, to the detriment of the accounting professionals’ views, was a major milestone and it would continue to inspire later statements produced on accounting theory. Obviously, new concepts meant new challenges and new issues to address, such as: how to link users’ needs and accounting principles; or how to deal with accounting users, a complex and heterogeneous group, both in nature and also in terms of needs.

Despite all efforts made by AICPA on accounting theory development, it failed to produce any statement in the 1960’s, and therefore pressure was piling on the APB. In response, the APB would finally publish the APS 4, Basic Concepts and Accounting Principles Underlying Financial Statements of Business Enterprises (American Institute of Certified Public Accountants; Accounting Principles Board, 1970a). The APB statement kept the focus on the objectives and on the user approach that has been used previously in the ASOBAT, as clearly stated in paragraph 9, and in paragraph 10, respectively:

“Accounting is a service activity. Its function is to provide quantitative information, primarily financial in nature, about economic entities that is intended to be useful in making economic decisions.”,

“Financial statements are the means by which the information accumulated and processed in financial accounting is periodically communicated to those who use it. They are designed to serve the needs of a variety of users, particularly owners and creditors.”.

The APB statement also focused on the accounting environment and the way it was linked to the users’ needs (paragraph 18, APS 4):

“Needs and expectations of users of financial statements are a part of the environment that determines the type of information required of financial accounting.”, as it also devoted a chapter to the links between the objectives of financial accounting and its users needs, in the scope of the accounting principles:

“The basic purpose of financial accounting and financial statements is to provide financial information about individual business enterprises that is useful in making economic decisions (…). General and qualitative objectives aid in fulfilling this basic purpose and provide means for evaluating present and proposed accounting principles.”.

By “qualitative objectives”, the APB meant that information needed to be relevant, understandable, verifiable, neutral, timely, comparable, and complete.22

Also of special interest, are the APB’s noteworthy views on GAAP (paragraphs 138 and 139, APS 4):

22 Previously, in AAA’s ASOBAT, some of these “qualitative objectives” were identified as standards - or criteria - of financial information, that would need to conform to a set of guidelines. More precisely, as standards, the AAA identified: relevance, verifiability, freedom from bias, and quantifiability. According to Hendriksen & van Breda (1992), the definition of standards was an attempt to avoid the problem of the myriad of different needs from a heterogeneous group of users, by producing financial information that was broad in its characteristics.

“Generally accepted accounting principles therefore is a technical term in financial accounting. Generally accepted accounting principles encompass the conventions, rules, and procedures necessary to define accepted accounting practice at a particular time. The standard23 of "generally accepted accounting principles" includes not only broad guidelines of general application, but also detailed practices and procedures.”24

“Generally accepted accounting principles are conventional - that is, they become generally accepted by agreement (often tacit agreement) rather than by formal deviation from a set of postulates or basic concepts. The principles have developed on the basis of experience, reason, custom, usage, and, to a significant extent, practical necessity.”.

Although the tacit nature of GAAP is recognized, the way and the objective foundations for its establishment are not explained. Despite the importance of APS 4, a prolific document regarding accounting theory substance, the GAAP definition is a good example of the overall APS normative approach. In fact, as the APB recognizes in the statement’ body text itself, the APS 4 is “primarily descriptive and not prescriptive” (paragraph 3, APS 4). Additionally, it also recognises the shortfall of discussion, and admits that new developments are missing, as the statement simply “identifies and organizes ideas that for the most part are already accepted.” (paragraph 3, APS 4). Finally, the AICPA’s APB was not concerned at all with enforcing the statement. On the contrary, it removed any responsibilities by unequivocally disclaiming that (paragraph 4, APS 4):

23 “The independent auditor's report gives the auditor's opinion as to whether the financial statements present fairly the financial position . . . and the results of . . . operations, in conformity with generally accepted accounting principles. . . .” (American Institute of Certified Public Accountants; Accounting Principles Board, 1970a). 24 “The term principles of accounting as used in reporting standards is construed to include not only accounting principles and practices but also the methods of applying them.” (Statements on Auditing Procedure No. 33, Auditing Standards and Procedures, p. 40, op. cit. American Institute of Certified Public Accountants; Accounting Principles Board, 1970a).

“The Board has not evaluated or approved present generally accepted accounting principles except to the extent that principles have been adopted in Board Opinions. Publication of the Statement does not constitute approval by the Board of accounting principles that are not covered in its Opinions.”.25

Unsurprisingly, although APS 4’s importance was recognized by the accounting community, criticism could not be precluded, even if such criticism has been much more limited than the one rose before by the ARS 1 and the ARS 3.26 As Hendriksen & van Breda (1992: 109) note, APS 4 “is not a theory of accounting practice nor a clear statement of GAAP”. Moreover, AICPA’s APB cannot be much criticized for attempting to be successful where others previously failed, where any attempts are likely to fail: undoubtedly, accounting theory is complex and diverse, and therefore its discussion is unlikely to result in any broad consensus. Moreover, pragmatism was increasingly overshadowing theorisation, and business lobbying and political influences were increasingly overruling the accounting profession intents of leading the development of accounting practice under a sound and broad theoretical framework that could effectively protect accountants’ conduct from such forces (Hendriksen & van Breda, 1992).

The AICPA’s APB lifetime was far from easy, as it had to face permanent criticism up to the point where many believed that the SEC would replace the APB in the role of the accounting policy setter (see e.g. Anthony, 1963). The quest for accounting principles required an adequate organisation, and both the AAA and the AICPA made proposals concerning how accounting principles and financial accounting objectives should be set. As to be discussed in chapter 2, it

25 Original italic. 26 For critiques and additional comments on the APS 4, see e.g., Ijiri (1971), Schattke (1972), or Staubus (1972). One of the members of the APB, George R. Catlett, dissented to APS 4. According to Catlett, the statement failed to provide “a basis for guiding the future development of financial accounting”, as the APB “is looking backward to what has occurred rather than forward to what is needed” (APS 4, 1970). Among other considerations, Catlett summarised his fundaments for dissenting by claiming that APS 4 “creates a significant roadblock which will seriously impede the efforts of the business community and the accounting profession to establish sound principles for financial accounting and reporting.” (APS 4, 1970).

28 is interesting to refer that one of the most controversial topics on APB’s agenda was the business combinations accounting. The discussion of this very sensitive issue would result in further pressure over an increasingly fragile board.

In 1971, under the AICPA auspices, a committee to study the establishment of accounting principles was created. The committee was under the leadership of Francis Wheat, and it would be called accordingly as the Wheat Study Group. Its report of the study on establishment of accounting principles, Establishing Financial Accounting Standards, was published in 1972. The Wheat Committee Report documented and analysed the difficulty of the search for accounting principles and, following the developments in the scope of the AICPA’s APB restructuring that was taking place by then, made two foremost recommendations: first, to replace the designation “principles” by “standards”, as “accounting principles” was considered slightly pretentious; secondly, to name the new standard setting organisation “accounting standards board” and not “accounting principles board” (Hendriksen & van Breda, 1992: 109).

The Wheat Committee recommendations led to the foundation of the Financial Accounting Foundation (FAF), which was made responsible for the FASB and for the Financial Accounting Standards Advisory Council (FASAC). This new accounting organisational structure was created in 1973 and, from 1 July, the FASB replaced the APB in the standard-setting role.27 Three main characteristics of the new board are worth highlighting: i) independence from the AICPA although, together with other organisations, AICPA continued to participate in the new board; ii) standard-setting authority given by amendment of Rule 203, providing enforcement power to FASB’s standards across the several states in the USA, as recognized by the state Boards of Public

27 While the FASB and the FASAC are professional bodies, the FAF is a foundation composed by trustees from a broad range of interested organisations, namely: the AAA, the AICPA, the National Association of Accountants, the Financial Analysts Federation, the Financial Executives Institute, the Securities Industry Association, and several governmental accounting groups.

Accountancy.28 The SEC has also endorsed the standard-setting authority of the FASB in 1973, ensuring substantial authority support to FASB’s standards; and iii) a new comprehensive accounting standards approval due process: topics are subject to a preliminary evaluation, followed by a possible admission to the agenda.29 Early deliberations are made public in discussion memorandums, before tentative resolutions are published in the form of exposure drafts, which are then subject to public discussion. Following hearings, further deliberations are possible, before a final resolution may be published as a standard.

Following the Wheat Study Group recommendations, the newly-appointed board started to look for accounting objectives, but using different designations from its predecessors. A new shift on the quest for accounting principles occurred. From onwards, instead of accounting principles, the focus would be on

28 According to the Rules of Conduct of the Code of Professional Ethics of the American Institute of Certified Public Accountants, the Rule 203 (vid. American Institute of Certified Public Accountants, 1977):

“prohibits a member from expressing his opinion that financial statements are presented in conformity with generally accepted accounting principles if the statements depart in a material respect from such principles unless he can demonstrate that due to unusual circumstances application of the principles would result in misleading statements - in which case his report must describe the departure, its approximate effects, if practicable, and the reasons why compliance with the established principles would result in misleading statements.”.

Following the Rule 203 amendment, the preponderance of FASB’s GAAP was made clear (SFAC 1, 1978):

“Rule 203 prohibits a member of the American Institute of Certified Public Accountants from expressing an opinion that financial statements conform with generally accepted accounting principles if those statements contain a material departure from an accounting principle promulgated by the Financial Accounting Standards Board, unless the member can demonstrate that because of unusual circumstances the financial statements otherwise would have been misleading.”.

29 Securities and Exchange Commission (1973).

30 accounting standards, having the “new” accounting principles a different meaning from the “old” ones.

From Paton to Moonitz, and from the AAA to the AICPA, the quest for a comprehensive and coherent set of accounting theory seemed to have reached an end. Principles (e.g. Hatfield et al., 1938; Paton, 1936), postulates (e.g. Moonitz, 1961; Paton, 1922), standards (American Institute of Certified Public Accountants; (Wheat Study Group), 1972; Paton & Littleton, 1940): looking back, the different developments in accounting theory seem now more a question of semantic than substance. Nevertheless, the previous work on principles and postulates would be useful for the creation of an accounting conceptual framework, as FASB’s availing proved, being such enterprise to be examined in the next chapter.

1.3 Ontology, theoretical identification and theory validation

Accounting theories can be identified and verified in different ways (American Accounting Association, 1971; Hendriksen & van Breda, 1992). Accounting theory can be classified as: i) language, ii) reasoning, or iii) script. As language, it can be studied: as: (i) pragmatics, the effect of language; (ii) as semantics, the meaning of the language; and (iii) as syntactic, the logic of the language. As reasoning, it can be classified as: (i) deductive, where arguments come from generalisations to specifics; or vice versa, as (ii) inductive. Finally, as script, the theory can be: (i) descriptive, also called positive, as it studies the interactions of financial information with its users; versus (ii) prescriptive, also called normative, as it discusses the ways the financial information should be prepared and reported.

As for the main approaches to accounting theory, there are three which relate more closely to the issues described in this chapter: i) the behavioural approach, which focuses on the relevance of the information that is used for decision- -making, and on the behaviour of the different users of the financial information

31 which is being reported; ii) the structural approach, which focuses on the structure of the accounting system itself. It is a classical approach to deal with basic issues in accounting, such as the ones related with professional judgement, or the relations between the accounting profession and the management (e.g. the study of the reasons underlying the management choices on accounting); and iii) the economic approach, which studies the correspondence between accounting data and economic interpretations. Among the several economic approaches, the microeconomic approach is of particular interest, as it focuses on the impact of alternative reporting procedures on economic activities, and on measurements at the firm level. “Institutional” modern accounting theory approach is based on microeconomics, as this is the approach adopted by FASB.30

Usually, inductive theories are positive, while deductive theories are not necessarily normative. The inductive reasoning is frequently based on experiments, and therefore it is often called empirical. The accounting theory is considered empirical, when, for instance, financial data is collected in order to infer, i.e. to induce, conclusions to a greater whole, the entire population, or the universe, at best. Nevertheless, it is important to avoid linear associations, as almost all theories include both inductive and deductive reasoning. Additionally, both inductive and deductive reasoning may be positive or normative.

As for theory as language in relation to accounting theories approaches, the structural approach is primarily syntactic, while the economic and behavioural approaches are primarily pragmatic.

In terms of theory verification, i.e. to verify whether a theory is logically sound and has a broad acceptance, prescriptive theories are verified by judging the reasonableness of their assumptions, while descriptive theories can be verified in two different ways: i) descriptive theories that do not include empirical content, like syntactic ones, are simply validated by logic; and ii) for descriptive theories

30 Classifications and definitions based on Hendriksen & van Breda (1992), as much of the explanations concerning this topic.

32 that have empirical content, like the pragmatic theory which focuses on the usefulness of accounting information to the users, its validation relies as much on its truth as on its usefulness.

The revolution in the accounting standard-setting process was also followed by a major shift in accounting research: from an inductive-deductive reasoning base, to a more pragmatic framework dominated by empirical studies.31 Pragmatism and usefulness become the standard for a substantial part of the accounting research, although with some disregard to other research strands that subsisted and continue to be developed.

It is a fact that dialectics enable social sciences to flourish, in the scope of the dynamic of the contradictions illustrated in the Hegelian logic – thesis, antithesis, and synthesis.32 According to Crowther (2002b: 8), the dialectics of corporate reporting is “concerned with the relationship between the organisation and its environment”, being mediated through financial reporting itself. However, the author argues that this corporate reporting dialectic is more apparent than real.

As descriptive theories gained ground, the number of tests enabling predictions has also increased. As long as the results of such predictions can be verified, a theory can be then validated. Such validation is achieved following a satisfying number of successful tests, i.e., theory can be validated through a certain number of repetitions. In case the theory fails to predict, or flaws are found, the theory is not confirmed, being ultimately refuted (Popper, 1963).

31 The “demand” for such change was materialised by Watts & Zimmerman (1978, 1979), whose theory would be later subject to further developments and consolidation (1986, 1990). According to Watts & Zimmerman (1990: 148), early examples of positive accounting research include Gordon (1964), Gordon et al. (1966), and Gagnon (1967). 32 Even though Kant (1781) was the first to develop the concept of dialectics, Hegel (1822) is considered the founder of modern dialectics.

One of the authors that focused on the description of the scientific process was Kuhn (1962). He suggested that science derives from paradigms, which are not necessarily to be accepted or refused by the scientific community, but are rather to be adopted according to its usefulness to generate pertinent questions about our world. If a paradigm is no longer useful to raise questions, it will then be abandoned.

Theory validation is everything but a consensual topic. While theory can be validated in different ways, some do not believe this to be essential, and others even argue that theory validation is not possible in social sciences. Naturally, the complex nature of accounting makes it more difficult to use predictions in order to validate theories.33

As this research is focused on the possible economic consequences of a particular change in accounting regulation by the standard-setting authority, an economic approach, primarily pragmatic, based on the study of the impact of the limitation of alternative reporting procedures on economic activities, namely on M&A activity, was adopted. This thesis is also based on descriptive theory, commonly referred to as positive, as it studies the interactions of financial reporting with the economic decisions of its users, rather than to be concerned in formulating ways in which accounting changes should be made, in the scope of the normative theory.

Both deductive and inductive theory are used, albeit the main reasoning is inductive, empirical, as several sets of financial data were collected in order to infer conclusions to a greater whole. As methodology involved the use of statistical testing, it was also possible to proceed with theory validation, with no disregard to the limitations referred before (Hendriksen & van Breda, 1992).

It is believed that the ontological and theoretical positions adopted in this thesis may allow the use of an adequate research methodology in accordance to the

33 Vid. Hendriksen & van Breda (1992: 20), for several levels of theory confirmation in accounting.

34 proposed examination of the main research question. It is also arguable that such approach may provide enough and suitable means to test the research hypotheses, in order to obtain results that may be adequately subject to validation. Finally, such adoption was also driven by the ontological-theoretical and methodological mainstream used in literature related to this thesis, as to be discussed thoroughly in this thesis.

1.4 Accounting, finance, and globalisation

Much of the historical and epistemological review presented in this chapter not only is suitable for supporting the accounting framework, but also for economics and finance. Financial accounting is placed at the micro-level of economy, and it is interrelated with finance. The approximation of financial accounting to finance is also a result of the exchange markets development and the changes in the accounting regulation. Additionally, not only accounting, finance, and economics share the same environment, as they also have the same scientific nature and basis.

Every discipline has its own particularities and therefore social sciences share as much similarities as differences. As discussed, the instrumental nature of accounting makes it special. For example, unlike economy, it does not make so much sense to simplify assumptions in order to formulate models from the reality, as accounting needs to find solutions for concrete solicitations from the real world. Concurrently, finance faced numerous developments in recent years, which have resulted in a flowing of new theories, deriving from innovative hypotheses that lead to the construction of new models. As examples of classical financial literature: the efficient-market hypothesis (Fama, 1970); the capital asset pricing model (CAPM) (Sharpe, 1964, 1970), in the scope of the portfolio theory (Markowitz, 1952); or the agency theory (Jensen & Meckling, 1976).34

34 For a history of corporate finance see e.g. Baskin & Miranti Jr. (1997).

In recent years, the paths of financial reporting and finance became increasingly linked. The research in financial accounting had to adapt to the developments in the securities markets and to the subsequent advances in the research in finance. Indeed, as the financial markets witnessed significant developments, the research in finance has flourished, propelled by a developing industry that was successful in attracting new researchers (vid. e.g. Mattessich, 1995). To a certain extent, such success was also made at the expense of the research in accounting. Indeed, as Mattessich (1995) notes, unlike finance, financial accounting is not particularly useful to the free market forces, but serves mostly institutions exercising regulatory functions. However, accounting research, unlike many other professional disciplines, such as finance, despite serving policymaking, does not lead practice, such service being construed as a disincentive for accounting practitioners to demand accounting research (Mattessich, 1995; Reiter & Williams, 2002). Unsurprisingly, the accounting research has been suffering from a long period of “crisis”, as proved by the weak demand for academic accountants by accounting firms, industrial firms, or by regulators; in contrast with a strong demand for academicians in other professional disciplines, such as finance (Demski et al., 1991; Reiter & Williams, 2002). Nevertheless, more recently, it became more common to FASB to try justifying their accounting policy options based on academic research on financial accounting and finance, as in the case of the intangible assets under business combinations accounting (Maines et al., 2003).

Concurrently to the evolutions and convulsions in the accounting field, a new strand of research, influenced by the advances in economics and finance empirical research, has emerged since the 1970’s (e.g. Watts & Zimmerman, 1978, 1986). This strand of research is focused primarily on the study of how economic decisions are made, rather than on proposing ways in which such decisions should be made, which was by the time of this insurgence the conventional approach in accounting (vid. e.g. Hendriksen & van Breda, 1992). Other strands of research in accounting have emerged since then, some of which as a response to this new strand, which is commonly labelled as positivist (inter alia Hendriksen & van Breda, 1992; Ryan et al., 2002). Nevertheless, the

36 positivist approach managed to establish as mainstream in financial accounting research, particularly in the USA.35 Consequently, since the 1980’s the accounting research published in the top journals was mostly based on USA data and has also employed a positivist approach (vid. e.g. Lukka & Kasanen, 1996). Nevertheless, it seems geography also plays a role: while researchers in the USA favour a more quantitative, positivist, approach; European researchers were found to favour a more qualitative (e.g. case studies) approach (Lukka & Kasanen, 1996), being some mainstream UK based journals a good example of prevalence of critical and interpretative approaches (Lowe & Locke, 2005). Considered by many as “revolutionary” (vid. e.g. Beaver, 1989; Mouck, 1993), the introduction of positivism in accounting generated widespread discussion among the accounting research community, that continues until the present, although which much less importance, as other strands in accounting continued to emerge and develop, such as constructivism (vid. e.g. Quattrone, 2000).

As every approach, positivism stands on is own merits. Undoubtedly, its inception induced major changes in the accounting research. However, as every approach, positivism has also its own drawbacks, as pointed out by many authors.36 Moreover, the emergence of positivism was not enough to stall the overall ongoing decline of the financial accounting research (Reiter & Williams, 2002). Neither so critical accounting, or any other alternative approaches in accounting, as such decline is a consequence of a complex set of reasons as examined by many authors, such as Mattessich (1995). Nevertheless, regardless criticism, positivism and economic-based research established as a mainstream approach not only in finance and in the USA, but also in financial accounting and in many other locations worldwide.

35 Bricker & Previts (1990) refer to an American Accounting Association Committee report quoting that the published research using a “rigorous” method increased from 12% in 1963 to 86% in 1975, and to 98% in 1986. Undoubtedly, the “pace of the empirical research revolution was rapid” in the USA (Reiter & Williams, 2002: 582). 36 Criticism to positive accounting research comes from different grounds: from theoretical (vid. e.g. P. Armstrong, 1991; H. G. Hunt & Hogler, 1990; Tinker et al., 1982; Tinker & Puxty, 1995) and philosophical-logical (e.g. Hines, 1988; Lowe et al., 1983; Mouck, 1992; Whittington, 1987), to methodological (e.g. Abdel-Khalik et al., 1989; Holthausen & Leftwich, 1983).

Finally, the need for accounting harmonisation appeared on the scene following the increasing of the economic globalisation pace. Concurrently to the FASB’s foundation, the International Accounting Standards Committee (IASC) was created in 1973 with the main goal to develop accounting standards that could be adopted not only at domestic level, but also internationally. Since then, the International Accounting Standards Board (IASB), its successor, has recorded some relevant accomplishments, namely the endorsement of international accounting standards by the European Union (EU) countries in 2005, and the acceptance of its standards by over one hundred countries worldwide.37 The FASB and the IASB have also been working together in several convergence projects, including the joint development of a conceptual framework. The SEC has also outlined a roadmap for the implementation in the USA of IASB’s pronouncements, the International Financial Reporting Standards (IFRS).38 As a consequence of the latter developments referred above, nowadays much of the discussions regarding accounting issues are made at a Pan-National level.

1.5 Conclusions

The last decades have witnessed major developments in the accounting practice and in the financial activity, understandably leading to a significant evolution in the accounting and finance research. As examined in this chapter, in the USA, the quest for accounting principles, that could integrate a comprehensive theoretical set, generated broad discussions among the accounting community that prevail hitherto. Concurrently, finance and financial markets experimented a significant growth and the research in finance has flourished, propelled by a developing industry that was successful in attracting new researchers. This has also affected the research in financial accounting, and a new strand of research,

37 IABS’s official website at www.iasb.org provides an updated list of countries that have adopted IASB standards, and also additional information about IASB’s history and activities. 38 IASB’s GAAP are essentially formed by recently issued IFRS, and by earlier standards, named International Accounting Standards (IAS), issued by its predecessor, the IASC.

38 influenced by economics and finance, has emerged. This strand of research, focused primarily on the study of how economic decisions are made, granted a significant adherence among researchers in financial accounting, regardless the emergence of other strands of research since then, often in opposition to such positive views. Despite disputable, as is everything in research, one could argue that the sizeable amount of research following a positivist approach published in the journals considered to be of the highest reputation in accounting and finance, leaves little doubts about the prevalence of this approach in the research developed in more recent decades, being this phenomena particularly evident in the USA.

It can therefore be regarded as reasonable that this thesis stands on a positivist ontology and methodology. This is not simply because a positivist approach seemed to be more appropriate for the development and testing of the research hypotheses, but also due to the fact that this research is focused on the examination of the possible economic consequences from changes in accounting regulation in the M&A activity in the USA, being the USA the most active market in terms of M&A industry, and also being the country with the most prolific technical (articles, reports, government documents, etc) and academic production with this regard. This reinforces the mentioned approach standpoint: much of the literature used to support the development of this thesis shares a positivist view, therefore expanding research possibilities and enabling diverse research synergies, such as cross-sharing of theories, methodologies, and findings.

However, such a standpoint cannot, by any means, be regarded as being disrespectful to other approaches largely considered to be equally valuable and valid from a scientific standpoint. Indeed, the main research question of this thesis, recalled in the previous paragraph, is predisposed to interdisciplinarity, ranging from the economic and social fields up to motivational and the underlying psychological related factors. Understandably, a comprehensive coverage of the extension of such interdisciplinarity, which would involve the assessment of a wide variety of topics, related to economic, finance, behavioural

39 finance, financial reporting, taxation, strategy, and motivational issues, among others, would not be reasonable, as the examination of the research questions imply some level of specification. Therefore, following a preliminary analysis of the issues more closely related to the main research question, in the scope of the research proposal, it has been considered to be reasonable that a focus should be given to socio-economic factors, with a particular emphasis to the M&A phenomenon and related economic and finance theories, and also to related financial reporting issues. Accordingly, other approaches were found to be of interest, namely involving the use of critical perspectives in accounting, related to issues regarding regulation and business and political influences; the use of survey method; and financial reporting analysis, related to M&A disclosures.

In order to fulfil the research proposals, the structure of this thesis has been outlined as follows. Following the discussion started in this chapter about the lost quest for principles in accounting and the role of regulation in financial reporting, chapter 2 examines the effect of lobbying and political influences on standard-setting policy, with particular regard to business combinations accounting.

The discussion of the main topics related with the new business combinations accounting and M&A activity, including its underlying theories, is shown in chapter 3. Of particular interest is also the discussion of the economic consequences that arise from the standard setting process, in the scope of FASB’s conceptual framework and FASB’s business combinations standards. This discussion is critical as it is directly related with the examination of the main research question.

One can argue that in recent years the accounting theoretical idiosyncrasies have been overcome by the financial theory developments. Nevertheless, despite some remarkable financial theory achievements, finance still fails to provide many explanations. Some of such theoretical shortfalls were identified by Brealey et al. (2006), and were described as notable “ten unsolved problems in finance”. One of the unsolved problems in finance is exactly to explain why the

M&A occurs in waves. The insufficiencies of both accounting and finance theories posed a major challenge to the present research. However, the theoretical insufficiencies were compensated by a considerable amount of literature concerning M&A, which allowed the development of hypothesis according to the main research question, as presented in chapter 4.

Chapter 5 exhibits evidence about the managers’ perception about the relevance of the changes in M&A accounting, collected by means of a survey questionnaire. It also provides an examination of disclosures related to business combinations accounting from annual reports of constituent companies of Standards & Poor’s (S&P) 500 index.

The remainder of this thesis is outlined as follows. In chapter 6 is provided information about the data collected for the development of the main research models. Chapter 7 presents the development of the research models, which, although founded in several existing hypothesis and partial theories, presents particularities as it was specially designed for testing the particular hypotheses formulated in the present research. Alongside with the model fitting, are also shown the main results and related statistical validations. Such results were triangulated with the evidence shown in chapter 5, being the discussion and interpretation developed in chapter 8.

Finally, the conclusions and suggestions for further research are shown in chapter 9. Among other contributions to existing knowledge, which are to be introduced and discussed in later chapters, the research findings have implications for several existing national sets of GAAP which keep allowing a traditional form of business combinations accounting. As IASB pursued the changes in business combinations deliberated by FASB in the US, and as a very substantial number of countries use or allow IASB’s GAAP, this has resulted in a dual business combinations accounting for dozens of countries worldwide. This is the case of many countries in the European Union which have a dual accounting system, where a more conservative accounting, based on the aged fourth and the seventh directives of the European Council, subsists alongside

41 with modern IASB standards, which were recently endorsed by the EU in 2005. Therefore, the finding on whether the new GAAP had economic consequences sheds light on the interest of an update of business combinations accounting at the domestic accounting level, in the scope of a desirable harmonised international accounting.

Chapter 2 Political Nature of Accounting Standard Setting

and Developments on Business Combinations Accounting

2.1 Introduction

Alongside an ontological and epistemological discussion, the previous chapter offered an overview of early developments on accounting regulation, from which it was possible to conclude that in early times the accounting profession in the USA was mostly dominated by private institutions. However, government regulation was also present, and that was not exclusive to the 1930’s, when the government regulated the financial reporting for a brief period. Furthermore, one could argue that political influence since then has been perennial, explicitly or implicitly, which is to be examined throughout this chapter.

The standard-setting process is therefore examined, with particular emphasis on the most recent decades in the USA, which includes the continuation of the discussion of why it was not possible so far to construct a comprehensive conceptual theory for accounting. Hence, this chapter covers the quest for a conceptual framework for accounting, co-substantiated by FASB’s pragmatic developments on conceptual framework; the issues raised whenever accounting choice is in practice or being discussed; and the examination of the effects of lobbying and political influences on financial reporting regulation.

Finally, this chapter also includes an in-depth analysis of the lobbying undertaken with regard to business combinations accounting, which offers excellent examples of influences on the standard-setting policy from both political and corporate lobbying parties.

2.2 Towards a regulated conceptual framework for accounting

As referred to in Chapter 1, the accounting regulation started in the UK with the publication of the Joint Stock Companies Act in 1844. It was also in the UK, a few decades before, that the steam engine was invented, leading to a public steam railway age which would arrive in the USA shortly after. And regulation in the USA would begin exactly with the development of railways, following a period of frenetic construction and growth (Hendriksen & van Breda, 1992: 61). The previous chapter also referred to the accounting regulation in the USA, which was in charge of the AICPA from 1887 to 1934, and in charge of organisms related to AICPA’s from 1936 until 1973, when FASB appeared on stage. This prevalence of private-sector regulation led Hendriksen & van Breda (1992: 59) to note that:

“The period since 1887 might be called the “Century of the Certified Public Accountant” because it is this branch of the profession that has dominated accounting in the United States. So much so that many see public accountants and accountants as one and the same. This dominance is quite astonishing when one considers how long accounting has been in existence and that for most of those thousands of years it was performed by managerial accountants for managers.”.

Consequently, there was a dramatic shift, as accounting became increasingly more important for external users, in opposition to its traditional role, being the accounting profession developing in a “self-regulation” mode.

Nevertheless, in the same year the AICPA (AAPA) was founded, financial regulation also began in the railway sector under the Interstate Commerce

Commission (ICC), created by the Interstate Commerce Act of 1887. The purpose of ICC was to regulate railway companies’ practices, which were by then being accused of several market abuses, due to increasing economic power, often obtained through M&A deals (Burns, 1998).39 The legacy of ICC was noteworthy, and not only for mergers in the railroad industry, as it also served as a model for numerous latter regulatory initiatives, including the SEC foundation. As regulation gradually overtook self-administered professionals in the USA, naturally including the accounting profession, Hendriksen & van Breda (1992: 59) also highlighted governmental and other interferences in terms of regulation:

“This period could also be called the “Century of Financial Regulation”, because for the better part of 100 years the U.S. government has sought to control the flow of financial information. This is equally astonishing when one considers that for thousands of years accounting went completely unregulated. What makes it even more remarkable is that almost all the empirical evidence gathered by academics in the past 20 years indicates that regulation of financial markets is less necessary than regulation of the airline industry, the telephone industry, or the trucking industry, all of which have been deregulated in recent years!”.

Indeed, despite the deregulatory trends of the past decades, the financial reporting continued to be subject of many developments in terms of regulation and governmental interference. The process of building a conceptual framework for accounting constitutes an example that illustrates the increasing role of regulation during the 20th century.

The conceptual framework was intended to be an equivalent of a constitution, a large umbrella designed to encompass accounting objectives and practices. The

39 The first railway merger in the USA occurred in 1838, when “Philadelphia, Wilmington & Baltimore” was created from the merger of “Baltimore & Port Deposit” with “Wilmington & Susquehana” (Burns, 1998). The development of the railway industry was so fast in the USA, that between 1884 and 1888, 425 M&A deals were recorded, and by the turn of the century, about one-sixth of the railway firms in the USA had been involved in consolidations (Association of American Railroads, undated).

FASB recognised such desiderate in a document entitled Scope and Implications of the Conceptual Framework Project (1976b). Accounting standards would be under the scope of a conceptual framework, just like a legal system in which laws are under a constitution. However, the adoption of this type of juridical structure was also an indication of the forthcoming “political nature of accounting standard setting” (Hendriksen & van Breda, 1992: 123). It became evident that the accounting construction process, primarily based in the accounting profession, would never be the same again.

Unsurprisingly, the political nature of the standard setting process led to a wave of criticism, as some of the worst fears from the accounting community were made real.40 The way that the political process has influenced the standards setting process in the USA is widely documented in literature (see e.g. Watts, 1977; Watts & Zimmerman, 1978), being this topic to be subject of further examination throughout this chapter. The accounting profession seemed to have become a victim of itself, as its leaders (Hendriksen & van Breda, 1992: 83):

“politically averse to government regulation and believers in the virtues of private enterprise, found themselves drawn into widespread and pervasive self-regulation in order to avoid what they perceived to be the greater evil of external regulation”.

From a previous framework, from which GAAP were generated by accounting practice to a new era of increasingly tightening regulation, where standards tend “to be set reactively rather than proactively” (Hendriksen & van Breda, 1992:

40 Some of such fears would be confirmed sooner than expected with the Watergate scandal. According to Hendriksen & van Breda (1992: 82), in 1973, after FASB took charge of the standard setting task, it was discovered that 17 major companies had made illegal political contributions. This event would jeopardise FASB’s and even SEC’s creditability. Eventually, the US Congress investigation would blame the FASB, and would urge the SEC to rule a “framework of uniform accounting principles”. With the stakes at such level, the FASB had necessarily to act promptly in order to prepare a conceptual framework for accounting.

84).41 This fate would continue until today, and is very unlikely to change under the currently in force paradigms.42

The construction of the conceptual framework was done throughout several stages. An early work assigned by the AICPA on accounting objectives, was produced by a study group, called Trueblood Committee, after Robert M. Trueblood.

According to Zeff (2004: 13), the 1960’s had been a decade in which American companies and also the USA government, in some instances, “had lobbied insistently against the APB’s proposed Opinions”, leading to a wave of criticism. Such criticism would lead AICPA to establish the Wheat Study Group, as referred in Chapter 1, and also the Trueblood Study Group (Zeff, 2004). The Trueblood Committee’s report, Objectives of Financial Statements, published in 1973, was based on a great extent on APB’s APS 4. Based on this report, the FASB started immediately its own work on the conceptual framework.43 As a result, it would publish later, in 1976, a Discussion Memorandum entitled Conceptual Framework for Financial Accounting and Reporting: Elements of Financial Statements and Their Measurement.44 In that discussion memorandum, a hierarchy of elements in a conceptual framework for financial accounting reporting was set in accordance to the following top-bottom sequence: objectives, information needed, qualitative characteristics, fundamentals, standards, interpretations, and practices.

41 Vid. also pp. 246-257. 42 Although some authors, such as Watts (2006), prospect a nebulous fate for the future of the accounting profession and regulation. 43 Vid. paragraph 57, SFAC 1 (Financial Accounting Standards Board, 1978). 44 Financial Accounting Standards Board (1976a). The Board’s document raised a high public interest, as FASB “received 95 written communications responding to the Discussion Memorandum, and 20 parties presented their views orally and answered Board members' questions at the hearing” (paragraph 58, SFAC 1, Financial Accounting Standards Board, 1978).

FASB’s discussion memorandum preceded the first piece of the conceptual framework: the Statement of Financial Accounting Concepts No. 1 (SFAC 1), Objectives of Financial Reporting by Business Enterprises, issued in 1978.45 The SFAC 1 (Financial Accounting Standards Board, 1978), which was founded on the Trueblood Committee report, addressed the top layer of the conceptual framework hierarchy: the objectives of financial reporting.

Conceptually, financial reporting is usually referred to as a broad concept. For instance, Crowther (2002b: 1) epitomises the conventional role for corporate reporting in public companies as:

“it provides a means for the organisation, or its representatives, to communicate the past actions of the company, the results of those past actions, and the intended future actions of the company. (…) Depending upon the perspective one takes, this communication may be to the owners of the business (i.e. the shareholders), the investors in the business, prospective future investors in the business, or to any permutation or combination of stakeholders who are associated with the business in any way.”.

Nevertheless, as the author unveils four stages of evolution of corporate reporting in the UK, it is revealed that in a third stage of development, post- -1975, a dramatic change occurred (Crowther, 2002b: 9):

“No longer was the firm seeking to communicate internally – to members or potential members – but rather the focus was upon the external environment. (…) the emphasis changed to the future

45 The high public interest continued throughout the entire due diligence process (paragraph 62, SFAC 1, Financial Accounting Standards Board, 1978):

“The Board received 283 written communications on the subject of the August 1977 hearing, of which 214 commented on the objectives and 221 commented on the elements, and 27 parties presented their views orally and answered Board members’ questions at the hearing. The Board issued an Exposure Draft of a proposed Statement of Financial Accounting Concepts on "Objectives of Financial Reporting and Elements of Financial Statements of Business Enterprises," dated December 29, 1977 and received 135 letters of comment.”.

prospects of the business becoming more important than the past performance.”.

Such change occurred with the publication of The Corporate Report, a document published in 1975 by the UK Accounting Standards Steering Committee. The financial reporting was therefore suffering major changes in both sides of the Atlantic.

Concerning the financial reporting, the FASB (Highlights, SFAC 1, Financial Accounting Standards Board, 1978) understood it as:

“not an end in itself”, but as

“intended to provide information that is useful in making business and economic decisions”.46

As for the objectives of the financial reporting, they were summarised by FASB as (Highlights, SFAC 1, Financial Accounting Standards Board, 1978):

“ - Financial reporting should provide information that is useful to present and potential investors and creditors and other users in making rational investment, credit, and similar decisions. The information should be comprehensible to those who have a reasonable understanding of business and economic activities and are willing to study the information with reasonable diligence.

- Financial reporting should provide information to help present and potential investors and creditors and other users in assessing

46 On the one hand, the issue of the users’ economic decisions is related to the economic consequences from changes in GAAP. On the other, the issue of the economic consequences from changes in accounting regulation is critical for the present research, as this thesis examines the possible economic effects of GAAP changes on the M&A activity. It is therefore important to understand the role of the users in the standard-setting process, as a change in the regulation may influence a user decision on a M&A deal, particularly at the managerial level. The issue of the possible economic consequences from changes in accounting regulation is to be referred in this chapter, being discussed further in Chapter 3.

the amounts, timing, and uncertainty of prospective cash receipts from dividends or interest and the proceeds from the sale, redemption, or maturity of securities or loans. Since investors’ and creditors’ cash flows are related to enterprise cash flows, financial reporting should provide information to help investors, creditors, and others assess the amounts, timing, and uncertainty of prospective net cash inflows to the related enterprise.

- Financial reporting should provide information about the economic resources of an enterprise, the claims to those resources (obligations of the enterprise to transfer resources to other entities and owners’ equity), and the effects of transactions, events, and circumstances that change its resources and claims to those resources.”.

Following early indications, the FASB kept the view of a financial reporting that needs to be user-orientated, i.e. that needs to be useful for decision-making. As for the user’s hierarchy, “investors” and “creditors” were elected as the most important ones, although they are to be (Highlights, SFAC 1, Financial Accounting Standards Board, 1978):

“used broadly and include not only those who have or contemplate having a claim to enterprise resources but also those who advise or represent them.”.

The establishment of objectives by the FASB raised many questions and critiques. For example, some authors criticised it as missing originality, since objectives were merely carried from previous works, such as the Trueblood Committee (see e.g. Most & Winters, 1977). Among the questions raised, perhaps the most pertinent was the one pointing out the complexity of the users of the financial information. Ideally, the financial reporting should provide information relevant and useful for every user. However, the user’s needs vary depending on its own nature, so it is impossible to satisfy the whole universe of users, particularly when the financial reporting is user-orientated.47 Not

47 It seems a paradox that user-orientated information cannot truly satisfy most of its users. However, this is a natural outcome from the confrontation of FASB’s views versus users’ heterogeneity.

50 surprisingly, the FASB selected the potential investors and creditors as the primary group of users, corroborating ASOBAT’s primary focus on owners and creditors, and in faithful observance of the Securities Acts of 1933 and 1934. However, looking more carefully at FASB’s intentions, one can argue that the board is admitting that the primary user is the individual and small investor, which is the one most deprived of financial information, when it is claimed that (Highlights, SFAC 1, Financial Accounting Standards Board, 1978): 48

“The objectives stem primarily from the needs of external users who lack the authority to prescribe the information they want and must rely on information management communicates to them.”.

There was therefore a shift on the user’s focus, as creditors continued to lose relevance to investors. The stewardship and accountability to shareholders was assumed as the main concern of financial reporting. This was a natural outcome of the rising importance of the investor, a noticeable phenomenon since the 1930’s. It was also a consequence of an interference of politics in the accounting regulation. As Tinker et al. (1988) argue, accounting became a weapon in the social clash of wealth and income distribution, suggesting that the decision of electing a primary group of users has as much of political as of technical. It is argued that politics protect the public interest better than accountants, who are more likely to favour privileged groups, such as shareholders, or the banking industry.

Naturally, the establishment of a new era for accounting was subject to widespread criticism. The construction of a set of objectives orientated exclusively for external users, the primacy of the investor, and the influence of politics in the accounting regulatory-setting process, are indeed doubtful options. For example, one can wonder whether the investor primacy is appropriate; or whether the objectives of the key users can be extended to other

48 Vid. Hendriksen & van Breda (1992 :129) for a better understanding of the underlying rationale. As Hendriksen & van Breda (1992: 83) note, it is the time of the “people’s capitalism”, where “ordinary folk would be encouraged to invest in corporations by the provision of relatively copious financial information”.

51 stakeholders, such as the management, which some authors continue arguing that should be regarded as the primary user group (see e.g. Flegm, 1984).

The heterogeneity of the users of the accounting information was epitomised by Beaver (1989). He presented five main groups of constituencies in the financial reporting environment: investors, information intermediaries, regulators, management, and auditors. From each main group were then identified sets of subgroups. For example, for the main group of regulators, three subgroups were identified: FASB, SEC, and the US Congress; while for investors were identified: diversified vs. undiversified; active vs. passive; and professional vs. nonprofessional. It does not seem plausible to admit that different groups of users share the same common interests, as it does not seem plausible that subgroups may share common interests as well. Several questions may be raised. A passive investor who relies on financial intermediaries for investment decisions may reject the idea of financial information directed towards him. In another example, a diversified investor may not be interested in firm-specific information. In conclusion, Beaver concludes that users are less likely to be homogenous, and its heterogeneity carries great complexity and difficulties for regulators.

The FASB’s views on accounting theory are necessarily pragmatic. This is reflected not only in the users’ approach, but also in every aspect of the accounting regulation. The FASB’s position on GAAP is a good example (Highlights, SFAC 1, Financial Accounting Standards Board, 1978):

“Unlike a Statement of Financial Accounting Standards, a Statement of Financial Accounting Concepts does not establish generally accepted accounting principles.”.

This assertion makes all the previous discussions on GAAP seem meaningless. The GAAP are simply to be established by the issuance of financial accounting standards, according to the developments in economic and financial environment, and following the evolution of users’ needs, with particular evidence for the investor. Therefore, the conceptual framework serves as a basis

52 for continuing to develop the GAAP, but not for the development of the GAAP concept. In fact, it seems that the FASB was pleased simply to continue developing AICPA GAAP’s. Moreover, it seems evident that FASB was not interested in getting involved into further in depth conceptual discussions on GAAP, which revealed to be such a fracturing issue.

In order to complete the conceptual framework, in accordance with the hierarchy set in the 1976’s discussion memorandum, six more SFAC were issued by the FASB between 1978 and 2000.49 The SFAC 2 (Financial Accounting Standards Board, 1980b) addressed the topic of qualitative characteristics of the financial information, while the fundamentals were treated in SFAC 5 (Financial Accounting Standards Board, 1984) and SFAC 6 (Financial Accounting Standards Board, 1985b).50 Concerning the GAAP, they are essentially built upon standards, called Statement of Financial Accounting Standards (SFAS), and they may be accompanied by interpretations and technical bulletins.51

Despite the efforts of FASB, professional judgement upheld its critical importance in accounting. Regardless the infinite number of standards and interpretative documents that a standard setter may issue, professional judgment will always be a key factor to achieve the objectives of financial reporting. In face of the many hundreds of documents, and the thousands of

49 The current conceptual framework is composed by six statements, as SFAC 6 (1985b) superseded SFAC 3 (1980a). 50 According to FASB (Highlights, SFAC 5, Financial Accounting Standards Board, 1984):

“The fundamentals are the underlying concepts of financial accounting - concepts that guide the selection of transactions, events, and circumstances to be accounted for; their recognition and measurement; and the means of summarizing and communicating them to interested parties.”.

Although the conceptual framework does not indentify the “fundamentals”, they can be found in the APS 4 (e.g.: accounting entity, going concern, time periods). 51 The FASB kept several AICPA’s standards, called APB Opinions, either unchanged or amended. Some SFAS were issued to replace APB Opinions.

53 words that compose the current GAAP set, it is therefore not surprising that many argue this to be worthless work, and a waste of public resources. Among the critics, Seidler (1984) argued that is not possible to have conceptual frameworks in social sciences, while Gerboth (1987) claimed that the major mistake of the conceptual framework was the conviction that setting principles in advance would minimize later discussions on basic accounting issues.

It was also unsurprising that the pragmatic approach from FASB extended to other counterparts around the world, resulting in a shockwave for the accounting discipline as a whole. As Crowther (2002b: 23) notes:

“These practical uses of accounting, as a means of running a business, contrast with the search by some writers for a philosophical base for accounting and a place for it in a wider societal context.”.

Consequently, several authors suggested and even developed alternative accounting theories and different approaches, some of which with a sound philosophical basis (see e.g. American Accounting Association Committee on Concepts and Standards for External Financial Reports, 1977; Chambers, 1979; Danos, 1977; Hakansson, 1978; Laughlin, 1987; Mattessich, 1980; Morgan, 1988; Tinker, 1988; Wells, 1976, 1977; Whittington, 1980).52

Regardless all the alternative views, FASB kept its pragmatism, also becoming more susceptible for some pragmatic lobbying, consequently opening ways for regulatory pressures, as to be discussed further in this thesis (vid. e.g. Hendriksen & van Breda, 1992; Riper, 1994).

52 For a different reading, it is also recommendable Mattessich (1995), which presents some curious analogies, such as the “onion model of reality”; and interesting confrontations, such as the analysis of the American ‘‘positive accounting theory’’ view versus the British ‘‘critical- -interpretive’’ view of accounting.

2.3 Lobbying and political influences on standard setting

The building of a conceptual framework could have served to somewhat shield the accounting standard setting process from interferences, placing substantive aspects at the service of users’ needs and on top of preparers’ interests. However, this was not to be necessarily the case, as the standard-setters’ perceived interest of users’ needs often clash with the interest of the financial reporting preparers, consequently generating disagreements - in some cases loudly expressed - , as to be shown throughout this chapter.

It can be regarded as natural that the standard-setting process in the USA frequently involves compromising, as the FASB endeavours to create pronouncements that may gather support amidst parties with possible conflicting interests, such is often the case of financial reporting preparers versus users (vid. e.g. Hendriksen & van Breda, 1992; Schultz & Hollister, 2003). Indeed, financial reporting regulation in the USA is supposed to guarantee a proper open due process, as FASB is subject to rules of procedure that offer possibilities for the general public to judge and comment on accounting issues (Mezias & Chung, 1989). Nevertheless, the standard-setting policy does not rely necessarily on FASB only. Indeed, managers may try to influence the accounting standard-setting process directly, by writing comment letters to the FASB or by lobbying the SEC; but they may also act indirectly by lobbying the Congress in order to put pressure on the FASB or the SEC (Johnston & Jones, 2006: 196; Sutton, 1984). Unsurprisingly, the standard-setting process can therefore result contentious, as pressure may come not only from industries, professional associations and groups, and individual companies, but also from opposing forces such as the SEC, or mighty government agencies (vid. e.g. Schultz & Hollister, 2003).

As referred earlier, the US Congress delegated regulation on financial markets and financial reporting to the SEC that, in turn, would soon later delegate

55 financial reporting regulation to private accounting committees and bodies.53 54 Nevertheless, both Congress and the SEC would not resist interfering in several occasions. For example, AICPA’s APB Opinion No. 2, Accounting for the “Investment Credit” (American Institute of Certified Public Accountants; Accounting Principles Board, 1962a) was overturned by the Congress, which by law allowed a different method, subsequently conceded by the APB in Opinion No. 4, issued with exactly the same title of APB Opinion 2 (American Institute of Certified Public Accountants; Accounting Principles Board, 1964). This controversial rearrangement did not find the approval of several APB’s members, leading to some loud disagreements, as it was the case of Leonard Spacek, who did not contain his harsh criticism to such level of interference (American Institute of Certified Public Accountants; Accounting Principles Board, 1964):

(Mr. Spacek) “believes this Opinion illustrates the accounting profession’s complete failure in its responsibility to establish accounting principles that will provide reliable financial statements that are comparable among companies and industries, for use of the public in making personal investment decisions. He states there is no justification for sanctioning two contradictory practices to accommodate SEC and other regulatory bodies and some CPAs who have approved reporting the investment credit as, in effect, profit from acquisition rather than from use of property. This flouts Congress’ clear intent in granting the investment credit, "to reduce the net cost of acquiring depreciable property." Alternative procedures under this Opinion can increase by up to 25 per cent the earnings otherwise reported. In this Opinion and in SEC's

53 As FASB notes in its disclaimer (Financial Accounting Standards Board, 2007b: 1):

“The SEC has statutory authority to establish financial accounting and reporting standards for publicly held companies under the Securities Exchange Act of 1934. Throughout its history, however, the Commission’s policy has been to rely on the private sector for this function to the extent that the private sector demonstrates ability to fulfill the responsibility in the public interest.”.

54 For a theoretical discussion about delegation of regulatory authority from a public principal to a private agent regarding accounting regulation, vid. e.g. Zeff (1995), or Mattli & Büthe (2005).

stated position, Mr. Spacek finds no word of concern for the investor, to whose protection both CPAs and SEC supposedly are dedicated. He believes this Opinion approves accounting of the type that precipitated the 1929 financial crisis, and that history is being repeated by actions of the very authorities created to prevent such catastrophes. He feels this breakdown in safeguards created to protect investors has resulted from fragmentation of responsibility for establishing accounting principles, and the only remedy is to create a Federally established Court of Accounting Principles with a prescribed basis for its decisions; this court would be independent of the profession and regulatory commissions, and its decisions would be binding on all, thus rescuing investors from their present abandonment.”.

Nevertheless, such proposed Court of Accounting Principles, or likewise, was never created, and accounting principles theoretical conception continued to be overruled by pragmatism, often suffering from external interferences.55

AICPA’s successor was also to be subject to interferences. For example, FASB’s SFAS 19, Financial Accounting and Reporting by Oil and Gas Producing Companies (Financial Accounting Standards Board, 1977), issued by a narrow 4-3 vote, faced aggressive pressure from the oil industry against an imposed “successful efforts costing” method, leading the SEC to take action (vid. e.g. Zeff, 2007). Indeed, some fierce lobbying, particularly from small and medium- -sized (SME) oil and gas companies, demonstrated clear opposition to the SEC regarding the discontinuation of the “full costing” method, which would, in turn, be transmitted to members of the Congress and of the government, such as in the case of the Justice and Energy federal department (Staubus, 2003; Zeff, 2007).

Following 12 days of public hearings and 2,700 pages of transcripts, the SEC would overrule FASB’s standard on the matter by issuing the Accounting Series Release (ASR) No. 253, Adoption of Requirements for Financial Accounting and Reporting Practices for Oil and Gas Producing Activities (Securities and

55 For more information on the Court of Accounting Principles vid. e.g. Spacek (1958), Ketz (2003), or Zeff (2001).

Exchange Commission, 1978; vid. e.g. A. J. Smith, 1981), proposing a new method, the “reserve recognition accounting”, and maintaining the previous methods during a transition period (Gorton, 1991; Zeff, 2007). That would force FASB soon later to issue an embarrassing SFAS 25, Suspension of Certain Accounting Requirements for Oil and Gas Producing Companies, an amendment of FASB Statement No. 19 (Financial Accounting Standards Board, 1979), suspending SFAS 19’s effective date, a move regarded as a clear concession to the SEC (vid. e.g. Gorton, 1991; A. J. Smith, 1981; Zeff, 2007).

SFAS 19 was approved following a close voting, with dissidents justifying their decisions based on technical-substantive matters. In the case of SFAS 25 decision on SFAS 19 suspension, only two FASB board members dissented, but with some incisive criticism to the interferences made, as they have questioned the private-sector nature of the financial reporting regulation, and the role of FASB itself (Financial Accounting Standards Board, 1979):

“Mr. Mosso and Mr. March dissent because they think that this action at least partially abdicates the Board's standard-setting responsibility. By suspending FASB Statement No. 19, the Board passes an opportunity to significantly narrow the range of accounting alternatives in the oil and gas industry and steps aside while the federal government attempts to resolve the issues. They think the Board should have retained Statement No. 19 as the duly adopted private-sector accounting standard, with an exemption for those companies (registered or nonregistered) electing to use the alternative full cost method permitted under SEC-prescribed rules. Although they disagree with the SEC decision that fails to adopt a standard for uniform use, in their view the Board's failure to narrow the numerous alternatives available to nonregistered companies to at least the two methods that registered companies must follow is not in the public interest. The absence of effective FASB-adopted standards may limit the Board from any significant role in the maintenance of these standards, including the successful efforts method of Statement No. 19. Mr. March and Mr. Mosso also believe more affirmative action would have placed the Board in a better position to maintain an active presence in this critical area of national

concern and to reassert leadership in setting accounting standards for the oil and gas industry.”.

Nevertheless, private-technical dissented voices are hardly taken into consideration when government authority is in force. Indeed, later the SEC would support - i.e. request - FASB to develop a method following the requirements of ASR 253’s “reserve recognition accounting”, for which FASB duly issued SFAS 69, Disclosures about Oil and Gas Producing Activities, an Amendment of FASB Statements 19, 25, 33, and 39 (Financial Accounting Standards Board, 1982), covering the due guidance for its implementation. Like SFAS 19, it was a close voting again, a 4-3 decision. Nevertheless, this time only technical-substantive argumentation was expressed, as no criticism to external influences was reported in the standard’s body text (Financial Accounting Standards Board, 1982).

More recently, Congress threatened to intervene unless the FASB would provide some relief in respect to fair value accounting, which led current FASB Chairman, Robert H. Herz, to testify about mark-to-market accounting before the US House of Representatives Subcommittee on Capital Markets, Insurance, and Government Sponsored Enterprises (Financial Accounting Standards Board, 2009; Investors Technical Advisory Committee, 2009), following active lobbying from the American Bankers Association (e.g. Beattie, 2009).

It is the mark-to-market issue and the controversial topic of fair value adoption, which has become increasingly notorious in recent years. FASB’s new rules for investments in the 1990’s materialised in SFAS 115, Accounting for Certain Investments in Debt and Equity Securities, (Financial Accounting Standards Board, 1993) and meant an increased use of fair value, which did not benefit financials industry’ valuations (Beatty et al., 1996). Unsurprisingly, share prices reaction for bank holding companies during the period surrounding SFAS 115 adoption was not favourable (vid. e.g. Beatty et al., 1996). Nevertheless the outcome for banks could have been worse in case FASB would not compromise, even if less than in other occasions. Indeed, limited concessions were made by

FASB following the comment letters received after the ED publication. According to the analysis made by Beatty et al. (1996: 92):

“FASB’s final positions on the specific issues raised in the ED were consistent with the preferences stated by the respondents except on the issues of addressing liability accounting and making portfolio transfers at fair value.”.

Despite pressures from the SEC, and in spite of the active lobbying from banking and insurance industries, widely documented in the literature (e.g. Beattie, 2009; Schultz & Hollister, 2003), this time the main goals aimed from such influence would not be achieved, as with SFAS 115 financial industry had to report securities at fair value, meaning a higher likelihood of volatility in reported earnings and equity.

Fair value accounting continued to undergo further developments in recent years. With SFAS 157, Fair Value Measurements (Financial Accounting Standards Board, 2006), FASB’s fair value approach was reinforced. Together with SFAS 159, The Fair Value Option for Financial Assets and Financial Liabilities, Including an amendment of FASB Statement No. 115 (Financial Accounting Standards Board, 2007c), the FASB increased the possibilities of using fair value and reformed the accounting for investments, by amending SFAS 115.

However, following the current financial crisis, the financial industry lobbied again against fair value in order to achieve its adoption easing, arguing that during a period of such severe low valuations and low liquidity, resulting in a strong emergence of the so-called toxic financial assets, the use of fair value would result misrepresentative (e.g. Accountancy Age, 2008; Beattie, 2009; Investors Technical Advisory Committee, 2009; Katz & Westbrook, 2009). At the height of the financial crisis, in the autumn of 2008, Edward L. Yingling,

President and CEO of the American Bankers Association (ABA), was not retiring as he was influential by stating that (Katz & Westbrook, 2009):56

“Given the importance of this issue, the impact it has on the crisis in the financial markets, and the seeming inability of the FASB to address in a meaningful way the problems of using fair-value in dysfunctional markets, we believe it is necessary for the SEC to use its statutory authority to step in and override the guidance issued by FASB”.

56 According to the American Bankers Association (2009a):

“Founded in 1875 and based in Washington, DC, the American Bankers Association brings together banks of all sizes and charters into one association. ABA works to enhance the competitiveness of the nation's banking industry and strengthen America's economy and communities. Its members – the majority of which are banks with less than $125 million in assets – represent over 95 percent of the industry's $13.3 trillion in assets and employ over 2 million men and women.”.

Being ABA’s mission described as (American Bankers Association, 2009a):

“to serve its members by enhancing the role of financial services institutions as the preeminent providers of financial services. This mission is accomplished through federal legislative and regulatory activities, consumer education, research, and products and services that promote, educate, inform and support members.”,

with a recognised and clear lobbying purpose (American Bankers Association, 2009b):

“Speaking for the entire banking industry and its two million employees gives ABA a powerful voice - on Capitol Hill, with the Regulatory agencies, in the courts and through the media. The resources in this area will help you keep abreast of the current GR priorities, as well as provide information on important issues, recent legislative or regulatory updates and key staff contacts within ABA.”.

Therefore the FASB and the fair value accounting become under fire, leading the Congress to threaten to take away standard-setting policy from the FASB, and forcing FASB chairman to take action. An excerpt from the 12 March 2009 hearing of a House Financial Services subcommittee follows (Investors Technical Advisory Committee, 2009; Katz & Westbrook, 2009):

“You do understand the message that we’re sending? panel chairman Paul Kanjorski, a Pennsylvania Democrat, asked Herz.

Yes, I absolutely do, sir, Herz replied.

Republican Ackerman followed Chairman Kanjorski’s comment, stating: I think what the Chairman said is if you don’t act we will.

After hesitating, Herz said he would try to get a new fair value rule finished within three weeks.

The financial institutions and their trade groups have been lobbying heavily, Herz said in an interview after the hearing.

Investors don’t lobby heavily”, he added.

Four days after such clear pressure, the FASB proposed an overhaul of fair value accounting, that according to Bloomberg could improve banks profits by more than 20 percent (Katz & Westbrook, 2009).

This evocation from FASB’s chairman of lobbying, made as a result of a clear political pressure, may resemble an embarrassing admission. Far from true, far from new, one could argue. For example, Fogarty et al. (1994), refer that earlier FASB leaders explicitly recognised political interferences in standard- -setting policy (vid. e.g. M. Armstrong, 1977; Kirk, 1978; Wyatt, 1986). However, Fogarty et al. (1994: 25) also note that admitting that accounting is political it does not necessarily result in the acceptance of such a situation (Solomons, 1978, 1991). Furthermore, Fogarty et al. (1994: 25) pinpoint that many accounting professional leaders believe that standard setting should be

62 insulated from politics (e.g. M. Armstrong, 1977; Kirk, 1978, 1986; Wyatt, 1986, 1990). In terms of realpolitik, anecdotal evidence suggests that standard-setting accounting policy continues to be influenced by corporate lobbying activity, which often results in pressures on the SEC and, whether necessary, even directly on the Congress and governmental agencies (vid. e.g. Beresford, 1989, 2001; Hinchman, 2000; Johnston & Jones, 2006).

As Herz admitted, preparers are indeed more willing and prepared for lobbying than investors, who are the weakest link in the equation. In the fair value case, the giant ABA met the Congress for another convenient alignment. Nonetheless, it is a fact that such interference could not avoid criticism and reactions from notable individuals and private associations.

Indeed, several individual and investor-advocate groups, opposed to this kind of arrangements that would enable banks to put off reporting losses, by allowing to use internal valuation models instead of market prices (Katz & Westbrook, 2009). For example, former SEC chairman Arthur Levitt, qualified them as outrageous threats (Katz & Westbrook, 2009):

“What disturbs me most about the FASB action is they appear to be bowing to outrageous threats from members of Congress who are beholden to corporate supporters”.

Furthermore, Conrad Hewitt, a former chief accountant at the SEC, meanwhile recognised that representatives from the ABA, American International Group Inc. (AIG), Fannie Mae, and Freddie Mac, have all lobbied him since 2007 in order to obtain the suspension of fair value accounting (Katz & Westbrook, 2009).57

Investors versus business interests or a David versus Goliath fight. Nevertheless, it is relevant to be aware that David has also an official voice within FASB, namely through the Investors Technical Advisory Committee (ITAC). The

57 For more pieces of evidence about lobbying on this issue, vid. e.g. Katz & Westbrook (2009).

ITAC is an advisory panel of investors that was established by the FASB to act as a barometer for investors relying on accounting rules, whose members include, among others, representatives from the CFA Centre for Financial Market Integrity, and analysts from Moody’s Corp, Standard & Poor’s, JP Morgan Securities, and from the California Public Employees’ Retirement System (Accountancy Age, 2009; Investors Technical Advisory Committee, 2009).58

On 15 June 2009, the ITAC, which, obviously, advocated that fair value accounting should be required for all financial instruments regardless the financial crisis conjuncture, sent an Unsolicited Comment Letter to FASB’s trustee, the FAF, with a set of proposals intended to help establish and maintain FASB’s independence (Investors Technical Advisory Committee, 2009). In this letter, the ITAC accused the FASB of being losing its critical independence of action (Investors Technical Advisory Committee, 2009: 2):

“The purpose of this letter is to share with the Trustees our grave concerns about what we believe to be a substantial erosion in the independence of the accounting standard setting process as

58 According to the ITAC (Investors Technical Advisory Committee, 2009: 1):

“The ITAC was established by the FASB for the stated purpose of serving as a standing resource to the Board and FASB staff that provides technical accounting advice, from the investors’ perspective, on current projects. The ITAC also identifies critical accounting and financial reporting deficiencies that require the Board’s attention and proposes new items to be added to the Board’s agenda, both major projects and technical application and implementation activities. The ITAC provides investor perspectives on the implementation of new standards.

The ITAC membership comprises thirteen investment professionals who are technically proficient in financial reporting and experienced in the analysis and use of financial statements. The members serve on an unpaid basis and have committed to devote significant amounts of their time and resources to assist the FASB in its mission to enhance financial reporting.”.

evidenced by recent events and a recent weakening of already inadequate accounting standards.”, therefore demanding necessary action:

“Given the disturbing potential for certain weaknesses to undermine the quality of the standards issued by the Board going forward, and further erode investor trust and confidence in financial reporting at a time of financial crisis partly attributable to poor transparency, we believe it is crucial that the Board and its trustee, the FAF, take swift and appropriate actions to protect and preserve the integrity of our financial reporting system.”.

A clear mention to the corporate and political pressures over fair value accounting changes was also made (Investors Technical Advisory Committee, 2009: 3):

“Even as these problems with the financial reporting infrastructure have become apparent, special interests and their agents, including some from the same financial institutions that were the original source of much of the current crisis, have worked to influence certain members of U.S. Congress to bring political pressure against the FASB and the U.S. Securities and Exchange Commission (“SEC”) to further weaken or even overturn the reporting standards for the trillions of dollars of “toxic” securities that led to the crisis.”,

To summarise, after decades of historical cost accounting dominance and debate, it now seems that fair value, or mark-to-market accounting, has become another major source of everlasting discussions focused in financial reporting, which seems to be increasingly subject to reporting volatility and to convenient regulation changes – also contributing in general for a more volatile reporting. Financial reporting also seems increasingly captive of corporate lobbying and

65 political pressures, propelling controversy among preparers, users, practitioners, and academics.59

The tracking record of political pressures and industry lobbying that the FASB and its predecessors have faced is a long one. Indeed, besides the topics referred, other accounting issues in recent decades, including leasing, or stock options accounting, led to all kind of pressures and active lobbying (vid. e.g. Dechow et al., 1996; Zeff, 2002). Whenever the FASB discusses a relevant financial reporting issue, that will necessarily generate reporting loses, one should expect a reaction from the aggrieved parties, which will likely result in to a help request from the interested parties to the SEC and/or to the Congress of the USA. In terms of theory of examination of corporate decision to resist FASB standards, Elbannan & McKinley (2006), developed a theoretical framework that specified the conditions under which corporations are likely to resist financial reporting standards, by examining three levels of determinants analysis were resistance occurs: the standard, the corporation, and the corporation’s industry.

As described throughout this section, lobbying and political influences on standard setting is a common phenomenon that can lead to pernicious effects. Schultz & Hollister (2003), using the case of the development of SFAS 115 to illustrate the pressures surrounding the standard-setting process, concluded that there is a tendency to make GAAP logically inconsistent; ensuring that some interested party is unhappy with the standards produced; ensuring changes in GAAP are inevitable, and also that it will be difficult to change standards quickly (P. W. Miller et al., 1998). Therefore, it appears to be the case that accounting regulation may also work to feed itself.60

59 For a recent discussion about pros and cons of fair value accounting from an academic perspective, vid. e.g. Laux & Leuz (2009) which, despite acknowledging difficulties of fair values implementation, do not regard historical cost accounting as a solution. 60 Also to be considered as potentially disturbing is whether financial reporting can be used as a political currency (vid. Ramanna & Roychowdhury, 2008).

It therefore seems that corporate and political pressures do play a critical role in financial reporting regulation, being business combinations a fine example of this type of interferences, as to be examined furthermore in this thesis.

2.4 Business combinations in the USA: an under pressure accounting issue

If we were to assess the amount of time spent on questions of pooling versus purchase, we’d find that this is the most costly accounting issue we’ve ever had in the US. Arthur Wyatt61

The history of business combinations accounting in the USA is necessarily short in time, as M&A lifetime only dates back to around 19th century, and, as discussed in chapter 1, regulated financial reporting is an even more recent phenomenon. However, despite this being a brief period of time, the history of business combinations is very rich and long in terms of accounting developments and events. Indeed, business combinations proved to be a long-running controversial accounting issue (Ribeiro & Crowther, 2006; Zeff, 2004). Supported primarily in Rayburn & Powers (1991), Wyatt (1963), and Carmichael et al. (1999), a brief review on some critical developments follows.62

The goodwill concept precedes in time business combinations accounting.63 However, what matters in the business combinations context is purchased goodwill, i.e., the difference between the price paid for completing the deal, less the net value of assets, deducted from liabilities, measured at fair value. Nevertheless, several evolutions in generic goodwill accounting are to be

61 Former chairman of Arthur Andersen and FASB member. 62 Besides this literature it is recommendable to consult G4+1 Position Paper (1998), particular its appendix, for a review and for a better understanding of more recent developments in business combinations and goodwill accounting. 63 For literature on early times in goodwill accounting see e.g. Dicksee & Tillyard (1906), Hatfield (1909), Yang (1927), Preinreich (1939), American Institute of Accountants; Committee on Accounting Procedure (1944), or R. H. Nelson (1953). For a more recent view on goodwill regulation see e.g. Holgate (1990), or Nobes (1992).

67 mentioned whenever considered appropriate for the understanding of the developments in business combinations accounting.

According to Wyatt (1963), in the early 20th century, business combinations were essentially considered as acquisitions, and therefore a single method was used to account for: the purchase acquisition. Wyatt (1963: 20) also notes that the accounting profession had long assumed as “GAAP” that:

“a new corporation could not begin operations with a balance in earned surplus. Or, stated more positively, earned surplus was presumed to have arisen from profitable operations of the entity on whose balance sheet it appeared.”.

Therefore, in a purchase acquisition, any retained earnings would not be simply carried forward to the new entity, as those were presumed to have resulted from profitable operations of the purchased entity. However, as Wyatt (1963: 20) refers, some business combinations that occurred in the 1920’s involved mere changes in the form of the entities, “without any real change in substance”, as illustrated bellow:

“For example, Company B and Company C, both subsidiaries of Company A, could be consolidated in a new Company D, or they could be merged, with the resultant entity being either Company B or Company C. Under these circumstances it appeared logical to carry forward the combined earned surpluses of the constituent companies to the new entity. This deviation from the general rule that a new corporation should not begin its existence with a balance in earned surplus was apparently well established by 1932.”.

Therefore, by the 1930’s it was starting to be widely accepted as a way to account for consolidations differently. And soon later, during the 1940’s, as the number of combinations completed and paid through an exchange of securities increased, the term “pooling of interests” made its first appearance. According to Wyatt (1963: 22), pooling of interests:

“was used to describe certain combinations in which the constituents had previously been so closely related that the existence of arm’s-length bargaining was questioned. In these cases the newly emerging entity wished to state its asset values at their fair value at the time of the combination, this value being measured by the value of the securities exchanged in the combination transaction.”.64

Despite the increasing use of pooling of interests, purchase accounting remained as the standard method for business combinations. In the early 1940’s, Paton (1943) considered the purchase acquisition as the “proper” method, while pooling usage was “questionable”.65 However, by the late 1940’s the high number of combinations completed through the exchange of securities, rather than payments in cash, led many accounting professionals to favour pooling of interests (Wyatt, 1963). The shift in payment for combinations, from cash to stock, meant that shareholders of the acquired firm would continue to own capital, but from the bidder, or in a newly created entity. Therefore, as it was no longer a sale of assets, it seemed logical to not revalue assets of the target entity.

The two methods of business combinations accounting become generally accepted in practice since World War II in the USA, but without being subject to any regulation. This gap was filled by the Accounting Research Bulletin (ARB) No. 40, Business Combinations, the first official pronouncement in business combinations, issued by AICPA’s Committee on Accounting Procedure (CAP) in 1950 (American Institute of Accountants; Committee on Accounting Procedure, 1950). This statement allowed the use of pooling if certain broad conditions could be met (based on a set of four criteria). Otherwise, purchase accounting was to be used. Pooling of interests was considered as a combination

64 The term pooling of interests was first used in a Federal Power Commission in 1943. Henceforth “pooling of interests”, “pooling of interests method”, and “pooling” will be used interchangeably in this thesis as terms with the same meaning. Idem for “purchase acquisition”, “purchase accounting”, “purchase method”, and “acquisition method”. 65 Op. cit Wyatt (1963: 21).

69 in which ownership interests continued, while purchase was understood as new ownership interests.

By then the use of pooling was not significant, as purchase accounting “did not necessarily result in charges against future reported earnings” (Walter, 1999: 39). However, with the publication of ARB No. 43, Restatement and Revision of Accounting Research Bulletins, issued in 1953 (American Institute of Accountants; Committee on Accounting Procedure, 1953), the CAP eliminated the option “to charge the excess of purchase price over the acquired firm’s book value to a capital account (called surplus)” (Wyatt, 1963, op. cit. Walter, 1999: 39). Consequently, acquirers could no longer avoid amortisation charges with negative effects in reported earnings. Naturally, pooling of interests became far more attractive than purchase accounting for bidding companies’ management.

As pooling’s popularity increased, the CAP felt compelled to tighten the criteria required for its use. Accordingly, it issued the ARB No. 48, Business Combinations, in 1957 (American Institute of Accountants; Committee on Accounting Procedure, 1957). The ARB 48 superseded ARB 43’s chapter in business combinations and revised the criteria for pooling qualification set in ARB 40. In terms of the main idea, according to ARB 48, a pooling of interests may take place when (paragraph 4, ARB 48):

“the holders of substantially all of the ownership interests in the constituent corporations become the owners of a single corporation which owns the assets and businesses of the constituent corporations, either directly or through one or more subsidiaries”.

Therefore, mergers and consolidations can continue to be considered for pooling of interests if certain conditions could be verified.66 In opposition (paragraph 3, ARB 48):

“a purchase may be described as a business combination of two or more corporations in which an important part of the ownership

66 Vid. paragraphs 4-6 of ARB 48 for pooling of interests requirements.

interests in the acquired corporation or corporations is eliminated or in which other factors requisite to a pooling of interests are not present.”.

As in ARB 40, the purchase method is more likely to be adopted for corporate acquisitions. However, the implementation of the refurbished requirements of ARB 48, which were supposed to tighten the requirements for pooling, proved to fail soon after. As Wyatt (1963: 61) notes:

“by the late 1950’s, the approach to the analysis of a combination transaction appeared to be that the absence of a given criterion should not prevent the transaction from being a pooling of interests if other features suggested that the treatment was appropriate.”.

In practical terms, pooling of interests continued to be a vague concept, and such lack of definition led to a loose interpretation of ARB 48 requirements (see also Sapienza, 1961). Consequently, both methods were widely used and during the 1960s pooling of interests became an even more popular method, certainly helped by the M&A wave which occurred in this decade (see e.g. Sapienza, 1962). More than 20,000 deals were completed between 1963 and 1969, about half of them at the peak 1968-69. However, what contributed more to the success of pooling was the possibility of an acquirer, suffering from declining profits, to trigger a financial performance turnaround through M&A with a company with steady or rising profits. As to be discussed in chapter 3, this period was epitomised by big conglomerates and diversification. Hostile takeovers were the fastest and preferred way to growth. And M&A deals could qualify for pooling in a relatively simple way. As Wyatt (1967: 211-212) recognises:

“the accounting for a combination is commonly decided in advance of consummation of the transaction”.

After some 20 years of standard-setting policy, CAP’s role was about to come to an end, amidst increasing criticism and allegations of incapacity of the

Committee to reach agreements in several accounting issues, such as changing prices, deferred taxes, pensions, and, has highlighted in this section, business combinations accounting (Zeff, 2001). The increasing use of pooling of interests and the failure of CAP to effectively constrain its adoption contributed to decrease its reputation and has increased the dissatisfaction with the pathway of business combinations accounting.

The CAP and the APB, or two sides of the same coin: the AICPA. Therefore it is not so much surprising that APB’s fate would be similar to its predecessor: failure to convincingly face criticism and disagreement with their standard- -setting policy. Rising opposition to the AICPA’s APB, which replaced the CAP in 1959, and criticism about the use of pooling of interests, led AICPA to publish two accounting research studies intended to discuss business combinations and goodwill accounting treatment: ARS 5 (Wyatt, 1963) and ARS 10 (Cattlet & Olson, 1968).67 Both these two AICPA’s ARS expressed opposition to pooling usage. Of particular importance, Wyatt’s study raised many doubts about the appropriateness of existing business combinations accounting, and about the use of pooling in particular (see e.g. Curran, 1964; Goldberg, 1964; Sapienza, 1964a, 1964b). Similarly, in its famous ASOBAT document, the AAA (1966) has also suggested to abolish pooling of interests. Many other voices would join the business combinations accounting discussion, primarily to express criticism to pooling (see e.g. Briloff, 1967, 1968; M. L. Davis, 1991; Eigen, 1965; L. C. Phillips, 1965; Rosenfeld & Rubin, 1985).68

Requests also started to be made to the APB to move forward in order to clarify the business combinations accounting issue (see e.g. Wyatt, 1965). However, the abolition of pooling was not favoured by many practitioners (see e.g. Holsen, 1963), and by the industry, as such a restriction would force to recognise goodwill, leading to a reduction in operational results, i.e., to a

67 ARS 5 (Wyatt, 1963) was quoted several times before in this section. 68 Several authors (e.g. Wyatt 1967: 211-212), reported abuses in pooling use, such is the case of retroactive pooling: “the nunc pro tunc change from purchase-type accounting to pooling-type accounting in recording a corporate merger” (Mosich, 1968: 352).

72 reduction of income without tax benefit (Sapienza, 1964a). Some were even so convinced about the need and superiority of pooling that argued that (Snavely, 1968: 88):

“The purchase method should, for the most part, be discontinued because it makes financial statement interpretation more difficult than necessary”.

The combination of a loose business combinations accounting and of professional interests, particularly from finance, resulted that pooling of interests became predominant in the second half of the 1960’s (Hong et al., 1978).69

Following ARS 5 and ARS 10 recommendations and related comments, and also the increasing wave of criticism over pooling, it was not possible any longer for APB to remain indifferent. Moreover, the visibility of pooling was increasing steadily, and so was its success. But pooling of interest’s fate was to become a victim of its own success.

Although it was urged to act, it was not possible to reach a consensus in AICPA’s APB about the selection between purchase and pooling of interests method; about how to capitalize goodwill and; if so, whether or not to amortise it (Carmichael et al., 1999). APB would finally issue an Exposure Draft (ED), Proposed APB Opinion, Business Combinations and Intangible Assets, in February (1970d), proposing to restrict substantially, and effectively, the use of pooling of interests; and to make mandatory amortisation of purchased goodwill, for a period up to 40 years, in opposition to write off directly against surplus.

APB’s proposals led to immediate negative reactions from academics, professionals, and from the industry (vid. e.g. Hendriksen & van Breda, 1992; Leftwich, 1981). For instance, Backman (1970: 47) gave an economic outlook on

69 Vid. e.g. figures from a 1955-58 study – 142 out of 263 mergers used pooling (Gagnon, 1971), versus the period 1966-67 – a study of 189 mergers revealed that pooling was used in 176 cases (R. M. Copeland & Wojdak, 1969).

73 the APB changes proposal and warned about the possible “bad economics” of the heavy burden of goodwill amortisation. At the same time he cast serious doubts about the APB preference of purchase method over pooling, as he believed purchase use would result in earnings being subject to inflationary forces. Wakefield (1970) cast an even worst prospect, suggesting that the new proposals would damage the M&A activity, resulting in “harm to the economy”.70 This type of argument would continue to be used throughout the lobbying history of goodwill and business combinations accounting.

Although most critics demanded discontinuation of pooling of interests, APB seemed forced to compromise and to maintain pooling of interests, even if in a restricted version. The discussions were long, and the board’s delay was resulting in additional pressure. Finally, under great pressure and broad contestation, including threats of legal action against, APB opted to maintain both methods and it published Opinion No. 16, Business Combinations in August 1970 (American Institute of Certified Public Accountants; Accounting Principles Board, 1970b).71 Simultaneously APB also issued Opinion No. 17, Intangible Assets, which confirmed capitalisation and amortisation of goodwill (American Institute of Certified Public Accountants; Accounting Principles Board, 1970c).72 As Zeff (2003a: 198) pointed out:

“the two Opinions were regarded more as the result of intense lobbying by industry than the product of sound thinking and analysis.”.

70 As an excellent example of dramatisation (Wakefield, 1970: 36):

“Consistency, comparability, and historic costs are the very foundations of modern day accounting practice. The new proposals will significantly undermine all three. The proposals of the Accounting Principles Board bring to mind the picture of a misguided Samson tearing down the pillars of his own house, under the impression that it is the house of Philistines”.

71 The effective date was for business combinations initiated after 31 October 1970. 72 The effective date was for intangible assets acquired after 31 October 1970.

Indeed, the decision on pooling of interests maintenance was everything but consensual among the board (American Institute of Certified Public Accountants; Accounting Principles Board, 1970b):

“Messrs. Davidson, Horngren, and Seidman dissent to the Opinion because it seeks to patch up some of the abuses of pooling. The real abuse is pooling itself. On that, the only answer is to eliminate pooling.”.

The possible impacts on M&A activity suggested from the industry were also dismissed by one the dissenting members of APB (American Institute of Certified Public Accountants; Accounting Principles Board, 1970b):

“Some say that to eliminate pooling will impede mergers. Mergers were prevalent before pooling, and will continue after. Accounting does not exist to aid or discourage mergers, but to account for them fairly. Elimination of pooling will remove the confusion that comes from the coexistence of pooling and purchase accounting. Above all, the elimination of pooling would remove an aberration in historical-cost accounting that permits an acquisition to be accounted for on the basis of the seller’s cost rather than the buyer’s cost of the assets obtained in a bargained exchange.”.

Despite all the criticism from many accounting experts and regulators, understandably not necessarily supported by the companies, APB managed somewhat to compromise, mitigating criticism from opponents of pooling by setting a set of 12 criteria requirement, to meet cumulatively, which constrained significantly the qualification of a deal for a pooling, enforcing the use of purchase method.73

73 Vid. the twelve conditions for pooling in APB Opinion 16 (Paragraphs 46-48).

Nevertheless, the fate of APB was set: it would be replaced by the FASB in 1973.74 Surprisingly, APB Opinion 16 and No. 17 would survive for more than thirty years, not necessarily untroubled. Immediately after the publication of APB Opinion 16, in a time-frame of around 3 years, the AICPA issued 39 Interpretations, and the SEC issued 3 ARSs, in order to provide additional guidance to the implementation of the pooling criteria (G4+1 (Organization) & International Accounting Standards Committee, 1998). The result: a progressive reduction in the number of companies qualifying for pooling of interests over time.

APB Opinion 17 required the recognition of intangible assets acquired in a business combination as well as the amortisation of purchased goodwill and other intangible assets over a period not to exceed 40 years. Consequently, a significant number of companies had to start recognising purchased goodwill: from less than 200 before 1970, to more than 400 afterwards, in a sample of 600 companies (vid. Fig. 2.1). The new GAAP have also resulted in a boom on the number of companies amortising purchased goodwill, which have become mandatory. Indeed, following the APB Opinion effectiveness, the majority of the 600 companies surveyed by AICPA (American Institute of Certified Public Accountants, 1968-2003) had goodwill amortisations: from less than 40 companies, to more than 250 few years after 1970. The intangible assets not being amortised shown in Fig. 2.1 refer to assets acquired prior to the effective date of the Opinion, and its figures become irrelevant as time passed, because

74 According to Carmichael et al. (1999: 1-32):

“The APB was weakened by nagging doubts about its independence, the inability of its part-time members to devote themselves entirely to the important problems confronting it, and the lack of coherence and logic of many of its pronouncements, which resulted from having to compromise too many opposing points of view”.

76 the amortisation periods of these pre-1970 intangible assets were progressively ending.75

No. of Companies 600

450

300

150

0 7 0 3 2 5 8 96 971 98 98 9 99 99 1 1 1974 1977 1 1 1986 1989 19 1 1 2001 Assets not being amortized Assets being amortized

Fig. 2.1 Goodwill recognised in business combinations, 1967-2001 Data source: based on a dynamic sample of 600 firms used by AICPA in Accounting Trends & Techniques, several ed. (American Institute of Certified Public Accountants, 1968-2003).

With the new business combinations accounting GAAP, the companies did not lose the pooling possibility, but now the odds of avoiding goodwill, minimising a negative impact on earnings, become noticeably reduced. This compromise, together with the replacement of AICPA’s APB by FASB led to some calm on the matter. However, the business combinations hot topic returned to the stage in the 1990s. The increase in M&A activity, with two M&A waves in late 1980s and in the 1990s, remarkable by their unprecedented magnitude, brought a renewed visibility to business combinations that would not escape FASB’s

75 AICPA did not provide information about the number of companies with assets not being amortised after 1991, presumably due to their insignificance.

77 notice. As FASB notes (paragraph B5, SFAS 141), the increase in M&A activity:

“brought greater attention to the fact that two transactions that are economically similar may be accounted for by different methods that produce dramatically different financial statement results. Consequently, both the representational faithfulness and the comparability of those financial statements suffer.”.

In addition (paragraph B6, SFAS 141):

“many perceived the differences in the pooling and purchase methods to have affected competition in markets for mergers and acquisitions. Entities that could not meet all of the conditions for applying the pooling method believed that they faced an unlevel playing field in competing for targets with entities that could apply that method.”.

The momentum offered an opportunity that had to be seised, as business combinations accounting presented many inconsistencies that had to be fixed in order to improve comparability not only in the USA, but also at international level. The topic proved to be a Pandora’s box in the past, well represented by the wave of criticism that AICPA’s APB suffered, and therefore such deep and strong criticism should no longer be ignored. Among the most important critics to take into consideration was APB decision to keeping accounting choice, by not simply discontinuing the use of pooling of interests (see e.g. Wyatt, 1972).76

76 APB Opinion 16 was approved by the minimum margin, as 6 members in 18 dissented (two- -thirds needed). Broeker, Burger, and Weston dissented because they believed “that, except for combinations of companies whose relative size is such as to indicate a significant sharing of ownership risks and benefits, business combinations represent the acquisition or purchase of one company by another and that accounting should reflect that fact.” (American Institute of Certified Public Accountants; Accounting Principles Board, 1970b) In a more radical approach, Davidson, Horngren, and Seidman dissented as they believed the Opinion to seek “to patch up some of the abuses of pooling. The real abuse is pooling itself. On that, the only answer is to eliminate pooling.” (American Institute of Certified Public Accountants; Accounting Principles Board, 1970b).

Another focus of contestation was the decision to require amortisation of goodwill and other intangible assets in a period of up to 40 years, ignoring assets having indeterminate lives as, for instance, perpetual franchises (American Institute of Certified Public Accountants; Accounting Principles Board, 1970c). This led 4 members of the Board to dissent, as they found such a requirement as much as inappropriate as arbitrary (American Institute of Certified Public Accountants; Accounting Principles Board, 1970c):77

“Whether amortization is appropriate depends on the particular circumstances of each case, including the evidence of increases or decreases in the value of such assets. In some cases, the facts may indicate maintenance or enhancement rather than diminution of value of the intangibles. In such cases, amortization is inappropriate. In other cases, the useful life may be determinable; then the cost should be amortised by systematic charges to income over the estimated period of usefulness. In all cases, the amortization of intangible assets should be used on professional judgment, rather than arbitrary rules.”.

In addition to the widespread criticism from practitioners, companies, and academicians, “arbitrary” and “inappropriate” are adjectives that some APB members used to describe key rules prescribed by the APB Opinions on business combinations accounting, for who maintaining pooling of interests method is no less than a “real abuse” (American Institute of Certified Public Accountants; Accounting Principles Board, 1970b, 1970c). Indeed, the continuation of pooling meant holding the accounting choice in business combinations accounting and also consequently a string of issues, including the possibility of further lobbying, as to be examined in the next section.

77 Burger, Davidson, Hellerson, and Horngren were 4 of the 5 members of APB that dissented on APB Opinion 17.

2.5 Accounting choice and business combinations accounting

As will be examined in this section, the existence of accounting choice constitutes a threat to the financial reporting credibility and often instigates lobbying actions. As referred earlier, the 1960’s were marked by persistent lobbying from the business community to APB’s proposals, to which the government of the USA would also adhere in some instances. One of the reasons for such active lobbying was related to the increasing awareness of managers concerning the importance of accounting choice. Moreover, the accounting choice was increasingly regarded as strategic in the case of M&A activity. As Zeff (2004: 13) notes:

“Company executives were awakening to the strategic importance of flexibility in the choice of accounting methods, especially when engineering, or defending against, company takeovers.”.

And it is a fact that inner issues at AICPA’s board made it unlikely to cope with lobbying pressures, tainting its deliberations (Zeff 2004: 13):

“Questions were raised whether a part-time board, such as the APB, could stand up against such pressures, because the accounting firms represented on the board had clients with vested interests in the outcome of the board’s deliberations. Many observers concluded that research had not contributed to a resolution of difficult accounting questions, as few of the APB’s Accounting Research Studies seemed to have an impact on the board’s thinking.”.

Together with APB’s insufficiencies, the controversy surrounding business combinations accounting proposals and final Opinions would also greatly contribute to its failure, as pinpointed by Zeff (2004: 13):

“Three Big Eight accounting firms are so critical of the intense ‘political’ lobbying of the APB leading up to Opinions 16 and 17 that they announce they have lost confidence in the APB as a source of sound financial reporting.”.

Indeed, the incapacity of APB to reduce accounting choice procedures undermined its credibility. As referred in paragraph 132 of SFAS 19 (Financial Accounting Standards Board, 1977):

“One of the principal criticisms of the work of the FASB’s predecessors that led to creation of the FASB was that they did not sufficiently narrow or eliminate free choice accounting alternatives. A report entitled Federal Regulation and Regulatory Reform (the "Moss Report") issued in 1976 by the Subcommittee on Oversight and Investigations of the U.S. House of Representatives and a report entitled The Accounting Establishment (the "Metcalf Report") prepared in 1976 by the staff of the Subcommittee on Reports, Accounting and Management of the U.S. Senate were both strongly critical of the availability of alternative accounting principles. In its November 1977 report, "Improving the Accountability of Publicly Owned Corporations and Their Auditors," Senator Metcalf’s Subcommittee concluded that "uniformity in the development and application of accounting standards must be a major goal of the standard-setting system."”

SFAS 19, which was referred earlier in this chapter, also offers specific arguments for the oil and gas industry that are also relevant for the broader spectrum of financial reporting (paragraph 132, Financial Accounting Standards Board, 1977):

“Moreover, two major financial statement user groups - the Financial Accounting Policy Committee of the Financial Analysts Federation (the national professional association of security analysts) and the Robert Morris Associates (the national professional association of bank lending officers) - have endorsed elimination of optional accounting alternatives not only for oil and gas producing companies but for other industries as well. The Securities and Exchange Commission, in Securities Act Release No. 5877 (October 26, 1977), took a similar position, stating that the Board's oil and gas project "is expected to result in significant improvement in financial reporting through the establishment of uniform accounting standards so that investors are provided with a valid basis for comparing the financial statements of different

companies." In the Board's judgment, when the same or similar facts and circumstances exist, as they do in the search for and development of oil and gas reserves, intercompany comparability requires a single method of accounting. Comparable reporting by companies competing for capital is, in the Board's judgment, in the public interest”.

This type of formal assertions justifying the superior interest of a single method for financial reporting’ users would be repeated in many occasions, as FASB tried to succeed where APB failed (vid. e.g. Financial Accounting Standards Board, 1985a, 1985c). The issue of business combinations accounting can be placed in such contexts, this time as a “second act” and with FASB as the protagonist, assuming therefore its examination a particular relevance for this thesis.

Previous to FASB’s final changes, accomplished in 2001, the accounting for business combinations was discretionary in the USA (American Institute of Certified Public Accountants; Accounting Principles Board, 1970b; Ribeiro & Crowther, 2007c). The issue of the discretionary nature of accounting choice has been subject to examination in the literature, often together with the analysis of lobbying effects. For example, research based on the positive accounting theory examined the determinants that instigate the business community and accounting firms to try to influence through lobbying the FASB, focusing in specific financial reporting pronouncements (Deakin, 1989; Kelly, 1985; Watts & Zimmerman, 1978, 1990).

Being aware that managers have discretion in the application of GAAP, the set of accounting procedures within which managers have such discretion can be called “accepted set” (Watts & Zimmerman, 1990). According to Watts & Zimmerman (1990: 136), the managerial discretion over the accounting method choice is expected to vary across firms with the variation in the costs and benefits of restrictions (enforced by external auditors) which will produce the “best” or “accepted” accounting principles. The managerial discretion can be used to meet financial reporting objectives. Moreover, the achievement of

82 objectives benefits managers whenever compensation is tied to accounting figures.

Assuming the economic view that individuals try to maximise their own utility, it is consequently admissible that “management lobbies on accounting standards based on its own self-interest”, being possible to examine predictors of lobbying behaviour (Watts & Zimmerman, 1978: 113). Watts & Zimmerman (1978), predict that corporate executives have greater incentives to lobby for accounting regulation that may result on increases in reported earnings, therefore increasing management wealth via, for example, increases in stock prices or stock options awards. As Watts & Zimmerman (1978: 118) note:

“In small, (i.e., low political costs) unregulated firms, we would expect that managers do have incentives to select accounting standards which report higher earnings, if the expected gain in incentive compensation is greater than the forgone expected tax consequences.”.

Indeed, higher earnings are desirable to managers linked to corporate reported performance, but related costs, such as taxes or political costs, also need to be weighed. Such costs may reduce stock prices and cash flows, and these negative impacts may outweigh the rewards that executives would have from an increase in reported earnings. Therefore, management may also lobby against accounting standards that result in an increase of reported earnings (Watts & Zimmerman, 1978: 118):

“managers have greater incentives to choose accounting standards which report lower earnings (thereby increasing cash flows, firm value, and their welfare) due to tax, political, and regulatory considerations than to choose accounting standards which report higher earnings and, thereby, increase their incentive compensation. However, this prediction is conditional upon the firm being regulated or subject to political pressure.”.

Subsequent accounting literature offers diverse motivations justifying why corporate executives may exercise accounting discretion among a set of

83 accounting procedures in order to obtain specific earnings goals (vid. e.g. Dechow & Skinner, 2000; Fields et al., 2001; Healy & Wahlen, 1999).

While examining the consequences of managers’ behaviour to shareholders’ wealth, Smith & Watts (1982) argued that the existence of bonus plans, tied to performance measured by accounting numbers, is justified as it allows to minimize agency costs. However, whether shareholders benefit from managerial discretion, and whether the benefits outweigh the costs is not such a clear matter (Fields et al., 2001). Additionally, Watts & Zimmerman (1978) found it reasonable to suggest that accounting choice may indeed affect shareholder’s wealth in case manager’s compensation contracts are tied to financial reporting performance.

The accounting choice has value implications, and this is evident in business combinations accounting.78 Unsurprisingly, the literature focused on the pooling- -purchase choice is large (Ashbaugh et al., 2004; Ribeiro & Crowther, 2007c). Several studies have found that both acquirer and target companies select a business accounting method based upon certain financial and non-financial characteristics (M. L. Davis, 1990). The percentage of insiders’ ownership, accounting-based compensation plans, leveraged-based lending agreements, the

78 Just like business combinations accounting, the valuation, in the sense of accounting measurement, is a controversial topic. What matters for the present research are the economic consequences of GAAP changes, and neither the valuation process itself nor its consequences. Therefore this topic is not treated in this research. For a theoretical views on accounting measurement see e.g. Ijiri (1967, 1975), Staubus (1985), or Hendriksen & van Breda (1992). In terms of discussion of assets measurement and valuation see e.g. American Accounting Association (1971), Penman (1970), or Sprouse (1971). For FASB’s views on measurement, vid. Financial Accounting Standards Board (1976a), SFAC 2 (1980b), and SFAC 7 (2000); and also SFAS 141 (2001a) and SFAS 142 (2001e) for measurement in business combinations accounting. As for approaches regarding measuring value for shareholders vid. Crowther (2002a), or Rappaport (1986), and also Cooper et al. (2001) for an empirical study on the subject. For a discussion on measuring and earnings manipulation see e.g. Dechow et al. (1995, 1996), Dechow & Skinner (2000), Lev (2003), or Louis (2004). Finally, for valuation in finance see e.g. T. Copeland et al. (2000), Cornell (1993), or Palepu et al. (2000), and for valuation and value creation in M&A see e.g. Gaughan (2002), Weston et al. (2004), or Sudarsanam (2003).

84 company size, and some other specific characteristics determine which business combinations accounting method is selected by the management (Dunne, 1990).79

Early studies found that managerial discretion is used in business combinations accounting in order to meet financial reporting objectives, namely to maximise reported earnings. As Copeland & Wojdak (1969: 188) note:80

“The manipulative quality of the purchase-pooling decision rule derives from the fact that acquired assets may be valued differently under the two methods. If a merger is accounted for as a purchase, acquired assets are recorded at the fair value of the consideration given by the acquiring company; however, under the pooling method they are valued at their preacquisition book values. (…) The method that minimizes asset values usually maximizes profits.”.

Unsurprisingly, diverse evidence supporting earnings management can be found in literature either in business combinations accounting (e.g. Anderson & Louderback III, 1975; R. M. Copeland & Wojdak, 1969; M. L. Davis, 1990; Erickson & Wang, 1999; Gagnon, 1967; Louis, 2004; Nathan, 1988), or in purchased goodwill accounting (see e.g. Jordan & Clark, 2004; Sevin & Schroeder, 2005).

Indeed, the issue of the earnings management in business combinations is not a recent one. Earlier than AICPA’s controversial publication of APB Opinion No. 16, Gagnon (1967) suggested that corporate managers may conveniently adopt one of the available business combinations accounting methods in order to best accomplish their particular goals, by maximising or smoothing the future reported income. Other authors, including Anderson & Louderback III (1975), and Nathan (1988), also present evidence suggesting earnings management in order to maximise reported income.

79 Vid. also Nathan & Dunne (1991) for some explanatory variables in purchase-pooling choice. 80 Also referring to Gagnon’s study (1967).

Additionally, Erickson & Wang (1999) and Louis (2004) found strong evidence suggesting that bidder companies overstate their reported earnings in the period previous to a M&A announcement, more precisely when involving a stock swap, in order to influence the exchange terms.

Evidence produced by more recent literature suggests that pooling of interests is in many cases the managers’ preference. In fact, research found that pooling of interest was preferred by managers, since the amortisation of goodwill by purchase firms was negatively perceived by the markets, as investors feared the dilution caused by goodwill (e.g. Ayers et al., 2002). In a more pervasive approach, it has also been found that managers from companies with compensations based upon earnings favour pooling because it benefited earnings and return on investment (see e.g. Aboody et al., 2000; Ayers et al., 2000, 2002).81

Evidence also suggests that such preferences for pooling of interests can be powerful enough to engage firms with M&A agreements’ provisions that would allow to cancel the deal in case pooling of interests could not be used (Aboody et al., 2000; Weber, 2004). Even more surprising is the fact that firms could be willing to pay in order to qualify for pooling of interests, as to be referred in this thesis.

2.6 Lobbying and pressures on FASB’s new M&A accounting proposals

As referred before, discussions about business combinations proposals and practices easily generate disagreements because they can lead to dramatic changes in financial reporting. Many companies and professional boards strongly disagreed with the pooling of interests discontinuation because they were concerned with managing cash flows and earnings per share (EPS) figures, and

81 Vid. also Gagnon’s (1967) seminal paper, which was yet to consider the role of managerial compensation contracts.

86 therefore they were afraid of compulsory goodwill recognition and amortisation charges (Aboody et al., 2000). They argued it would not be possible to reach completion of many M&A deals without the pooling option, being such difficulties particularly visible in major deals involving large companies with massive amounts of goodwill and other intangible assets. Therefore, according to the pooling elimination opponents, the M&A activity and the economy could suffer from this potential constraint, as to be examined in this section (see e.g. Briner & Fulkerson, 2001; E. E. Smith, 1999).

Without surprise, many respondents to the early FASB’s Invitation to Comment, Methods of Accounting for Business Combinations: Recommendations of the G4+1 for Achieving Convergence (Financial Accounting Standards Board, 1998), and to the first Exposure Draft, Business Combinations and Intangible Assets, released in September 1999 (Financial Accounting Standards Board, 1999a), argued that mergers should be accounted for differently than acquisitions (paragraph B36, SFAS 141). A few respondents supported that the pooling method should be applied to all mergers, while few others even recommended that it should be applied to all business combinations (paragraph B37, SFAS 141). These respondents mentioned different arguments to support pooling of interests, such as: more appropriate for “true mergers” or “mergers of equals”; or more reliable, more representationally faithful information, and more useful decision-making. However, not all respondents defended pooling continuation arguing, for example, that pooling elimination would enhance the comparability of financial statements.82 Regardless of the pro and cons views, the fact is that from over 200 comment letters received by the FASB and from the about 45 parties that testified at FASB public hearings in early 2000, some two-thirds of the commentators opposed to pooling of interests elimination and a large number were also against FASB’s proposals for goodwill treatment (Beresford, 2001: 74; Financial Accounting Standards Board, 2001a).

The Board refuted all the arguments of those who backed pooling, concluding that: i) M&A are similar in economic terms; ii) “true mergers” or “mergers of

82 SFAS 141, Basis for conclusions.

87 equals” are nonexistent or so rare as to be virtually nonexistent; iii) information provided by pooling is not a useful decision and less relevant in terms of completeness, predictive value, feedback value and comparability than the information provided by purchase method; and finally, iv) pooling is inconsistent with the historical-cost accounting model.83

Pooling of interests has been for a long an accounting option that benefited some industries, such as information technology (IT), telecommunications, and Banking and financial services, including companies such as Cisco Systems, AT&T, AOL, or American Express (vid. e.g. Beresford, 2001). It represented also an opportunity for creative accounting. However, the pooling benefits had inherent some relevant costs. Companies often consumed “substantial resources” structuring transactions merely to meet the requirements of pooling (Linsmeier et al., 1998; Lys & Vincent, 1995), spending massive fees with legal and financial advisors. In what would become a flagship illustration, Lys & Vincent (1995) collected evidence supporting the finding that AT&T paid a minimum of $50 million, and perhaps as much as $500 million, to satisfy pooling of interests accounting. Furthermore, this type of obsession could even lead to put the formal aspects over the corporate strategy, which is a mistake at M&A level (e.g. Reed et al., 2007).

Another reason for alarm was EPS, which is reported under SFAS No. 128, Earnings per Share (Financial Accounting Standards Board, 1997a). EPS is perhaps the most important financial performance indicator in the USA, as suggested by the importance given by companies in financial reporting and by the attention given by the financial mass media (e.g. Chant, 1980; Larcker, 2003; Marquardt & Wiedman, 2005).84 Companies concerned with EPS management tended to amortise goodwill over 40 years, therefore minimising the negative impact on earnings. Fig. 2.2 exhibits how companies dealt with the burden of goodwill amortisation until SFAS 142 replaced it by impairment tests. More than half of the companies surveyed chose large amortisation periods, two-

83 SFAS 141, Basis for conclusions. 84 Vid. Chapter 5 for more insights about EPS.

-thirds more than 20 years, and at least one third opted by the maximum dilution, the 40 years amortisation ceiling.

Other

Legal / estimated life

10-15

20 25-30

"Not exceeding 40"

Fig. 2.2 Average goodwill amortisation period in the USA, 1985-2001 Data source: based on a dynamic sample of 600 firms used by AICPA in Accounting Trends & Techniques, several ed. (American Institute of Certified Public Accountants, 1968-2003).American Institute of Certified Public Accountants, 1968-2003

Respondents’ views on the FASB proposals about goodwill and other intangible assets often varied.85 Some stated that no intangible assets should be amortised; others stated that a presumption about the length of the amortisation period was not necessary, nor was a maximum. Other argued that goodwill should be written off immediately because of the uncertainties associated with goodwill subsequent to its initial recognition. Various respondents favoured not amortising goodwill but testing it for impairment instead.

With the September 1999 FASB’s ED, the board proposed to limit the maximum amortisation period for goodwill to 20 years, presenting goodwill amortisation expense on a net-of-tax basis in the income statement, and not amortising only certain intangible assets. However, like APB thirty years before, the FASB would also be forced to compromise, abandoning this early proposal.

85 SFAS 142, Background information and basis for conclusions.

The reactions to FASB’s ED were of disagreement concerning pooling of interests elimination. (e.g Beresford, 2001). By that time the technological M&A wave was peaking (e.g. Thomson Financial - Sikora, 2002). Many of these deals involved the acquisition of companies with a high percentage of intangible assets. Furthermore, a bubble was forming in the exchange markets, particularly in the National Association of Securities Dealers Automated Quotations (NASDAQ), the stock exchange were most technological companies are listed. This resulted in M&A deals involving high premium bids, and therefore sizeable amounts of potential purchased goodwill and other intangible assets were to be recognised if purchase method was to be used. Naturally, pooling of interests seemed critical in many cases, as if these deals were to be accounted using purchase method, they would probably become suddenly uninteresting.86 Therefore ‘Silicon Valley’ was worried with the possible negative effects of pooling elimination on M&A activity and on its industry development (see e.g. King, 2000, Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000, Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000).

As FASB’s proposals on business combinations were unconvincing for corporate executives, and as FASB’s hearings were not being favourable for corporate lobbying, which continued to favour pooling of interests, the matter would then become a political one.

On 2 March 2000, Senator Phil Gramm, Chairman of the Senate Banking, Housing, and Urban Affairs Committee, held a hearing on ”The Pooling Method of Accounting for Corporate Mergers”, whose invitees would be grouped in two panels, an usual practice (Beresford, 2001). The first panel grouped prominent individuals from the business community, including American Express CEO

86 Examples of M&A deals that were carefully structured in order to qualify for pooling of interests includes the acquisition of NCR by AT&T, the megamerger of Daimler-Benz and Chrysler, and many other M&A deals, such as SBC Communications and Ameritech, Citicorp and Travelers Group, or Exxon and Mobil (Duncan & Carleton, 1999; Lys & Vincent, 1995; Walter, 1999).

Harvey Golub; a leading venture capitalist, John Doerr; Dennis Powell, Vice President and Controller of Cisco Systems; an James Barksdale, a partner from The Barksdale Group, among others (Beresford, 2001; Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000). The sympathy for this pressure group, unfavourable to pooling of interests elimination, was made clear in the opening statement of Senator Gramm, who referred to the first panel members as the “victims” (Beresford, 2001: 74). Moreover, the Senators declined to invite anyone supporting FASB’s ED, being its defence solely in charge of Edmund Jenkins, FASB Chairman at the time, and the single individual invited for the second Congressional panel (Beresford, 2001). It is also noteworthy that, according to Beresford (2001), FASB unsuccessfully proposed other individuals to testimony in favour of the board’s proposals on business combinations accounting.

In the first panel of the Senate Baking Committee Hearing on Pooling Accounting significant allegations were made about the dangers of FASB’s proposals, particularly concerning pooling elimination. James Barksdale, from The Barksdale Group, “a leading investment and advisory group committed exclusively to growing companies in Internet services” stated that (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000):

“I strongly disagree with the FASB’s plan to eliminate the pooling-of-interests method of accounting. I believe that the elimination of pooling will discourage mergers and acquisitions, will have a negative impact on the U.S. economy, and will create financial statements that are, at best, irrelevant and, at worst, misleading.”.

Not only was a clear opposition to pooling elimination expressed, but serious concerns were also expressed regarding negative impacts on the M&A activity and on the economy of the USA. Such allegations were detailed further (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000):

“The elimination of the pooling method will discourage mergers of knowledge-based companies and will have a chilling effect on the flow of venture capital to promising new sectors. Poolings have proven to be an effective and efficient way for companies to integrate new intellectual property and innovations into their operations. At the same time, the prospect of such combinations has provided incentives to the start-up companies that have driven much of the technology sector’s innovation. FASB’s proposed elimination of pooling will discourage such combinations, will make it harder to attract institutional capital and - ultimately - will dampen desirable business activity. Still worse, these negative impacts will occur without any improvement in financial statement reporting.”.

As possible negative consequences for innovation and R&D (research and development) for technological companies were highlighted - that the proposed changes in M&A accounting could only aggravate - James Barksdale also used its own experience as executive to argue that pooling was critical to undergo M&A deals in the information technology industry (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000):87

“From my own experience, I can tell you that the AOL/Netscape merger would not have occurred if pooling had not been an option. The bulk of Netscape’s value was intangible, and the amount of resulting goodwill that would have had to be amortized under the purchase method would have scuttled the deal. A report prepared recently by Merrill Lynch attempted to show the impact on AOL’s earnings if it had been required to use the purchase method in the Netscape transaction. If AOL and Netscape had used the purchase method, its 1998 loss would have been roughly ten times greater.”.

87 James Barksdale served as Chief Operating Officer of Federal Express, Chief Executive Officer of AT&T Wireless and most recently, and President and Chief Executive Officer of Netscape. At the time of the Senate hearing he was serving on the board of directors of a wide range of technology companies, including America Online, Sun Microsystems, Respond.com and Liberate Technologies (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000).

It was therefore suggested that pooling elimination would force any company possibly interested in a M&A deal to use an unsuitable purchase method that could prove rather unfeasible for a deal completion.88 It therefore seems clear that the proposed reduction of the purchased goodwill amortisation from 40 to 20 years was problematic - being its combination with pooling elimination regarded as possibly fatal for the M&A activity.

Understandably, criticism was made regarding a loss of quality of financial reporting whether proposed changes in M&A accounting would be effectively adopted (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000):

“To the extent analysts expand this trend of ignoring amortization charges, it will make the GAAP financial statements increasingly irrelevant for those companies that are large enough to be followed by analysts on a regular basis. This is compelling evidence that financial reporting will not be improved by FASB’s proposed change. In fact, financial statements will become less transparent. Analysts will ignore the drag on earnings caused by the use of purchase accounting in those transactions that are large enough to be closely followed by the market. For mid-sized or small companies, however, analysts and investors will not have sufficient information to enable them to look through intangible asset amortization. The same phenomenon is also likely to occur in the case of smaller transactions involving large companies. As a result, there will be different earnings benchmarks in the marketplace depending on the size of a transaction or the size of a company. Investors will not have the same type of cash earnings information or analysis for all companies. Similarly, requiring the use of purchase accounting will impair, not improve, comparability between the financial statements of companies that grow internally versus those that grow through business combinations.”.

88 Afterwards, the AOL/Time Warner deal would not qualify for pooling, as AOL Time Warner executives considered better to ignore the impact on financial statement reported earnings and, instead, focused on operating cash flow (Moehrle et al., 2001; Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000).

The statement above comprises serious allegations, namely a foreseen increasing irrelevance of FASB’s GAAP to firms followed by financial analysts; a reduction in the transparency of FASB’s GAAP adoption, as many companies are not subject to financial analysis evaluation; in resume, comparability would be impaired, not improved as argued by FASB. Direct references to the FASB were also made, regarding both the merits of its initiative but also the difficulties carried by its proposals (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000):

“FASB has identified an important issue - the need to improve accounting for intangibles. While there is a virtually uniform approach to valuing certain intangible assets such as cable systems and copyrights, consistent and well-tested valuation methods simply do not yet exist for many of the intangible assets that comprise today’s companies. For example, FASB has proposed that favorable government relations is a separately identifiable intangible asset. We have no guidance, however, as to how that asset should be identified, valued or lifed. As it stands now, if you asked five different valuation experts to value that intangible asset and determine its useful life and you would get five different answers. The lack of uniformity on these important issues will result in inconsistencies in the allocation of the purchase price between identifiable intangible assets and the residual goodwill between different companies. For this additional reason, consistency in financial reporting will not be improved.”.89

As the new M&A accounting involved also a revised treatment of purchased intangible assets, the implementation difficulties of the proposals were also used to highlight possible additional issues concerning a deterioration of financial reporting quality. It is also noteworthy that the example given to illustrate the difficulties regarding separately identifiable intangible assets valuation was “favourable government relations”. Can this also be considered a subtle form of pressure?

89 Original underline.

Arguments in favour of pooling of interests elimination were also refuted (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000):

“One final point deserves mention. Some have argued that pooling should be eliminated because it allows one company to overpay for another company or encourages companies to enter into unsound business combinations. Even where both sides to a merger want the deal to be completed, the parties do not have entirely common interests. As a result, the value of the consideration exchanged is, by definition, determined at arm’s length. In addition, corporate managers and directors have fiduciary responsibilities to their shareholders. To the extent those fiduciary duties are breached, the legal system, not the accounting system, should address those problems.”.

Finally, James Barksdale appealed for compromise and for a suitable solution for all parties regarding the business combinations issue (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000):

“I believe that we have an opportunity to develop a financial reporting model that provides accurate information to investors in the New Economy and that does not hinder the tremendous economic growth fueled by the technology industries. We should do so in a careful and thoughtful manner. There is no compelling rationale for the speedy elimination of pooling without thoughtful consideration of these larger issues. I look forward to a continued dialogue on these important issues and welcome your questions.”.

Pressure and lobbying on M&A accounting discussions were indeed evident in the USA, from the basic standard-setting level to the upper political level. These statements from James Barksdale were not alone in the hearing. Dennis Powell, Vice President and Corporate Controller of Cisco Systems Inc., remembered in the Senate hearing that he was with “a Big Five accounting firm for 26 years serving many technology companies, as well as having responsibility as a National SEC Reviewer”, as he prepared a background on business

95 combinations on his testimony (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):90

“The two methods of accounting - "Purchase" and "Pooling of Interests" - have been generally accepted in practice since 1945. In 1970, the Accounting Principles Board studied and discussed the pros and cons of the two accounting methods, and issued APB16 "Business Combinations", which concluded that they "find merit in the purchase and pooling of interests methods of accounting for business combinations and accepts neither method to the exclusion of the other." (APB16, paragraph 42). This viewpoint was reaffirmed in 1994 by a task force commissioned by the American Institute of Certified Public Accountants to study the usefulness of financial reporting. This report, entitled "Improving Business Reporting - A Customer Focus" concluded, after three years of study, that the existence of the two methods is not a significant impediment to users' analysis of financial statements and "A project to do away with either method would be very controversial, require a significant amount of FASB time and resources, and in the end is not likely to improve significantly the usefulness of financial statements." In rereading APB16, I found it interesting that arguments for and against the pooling versus purchase methods of accounting haven't changed over the past 30 years - we are still debating the same issues. However, while time has not changed the fundamental arguments for each method, time has changed the magnitude of the implications from the defects of the purchase method described in APB16, such as the difficulty in identifying and determining the fair value of intangible assets, including goodwill, and the

90 The simple presentation of the main facts of Cisco, as offered to the hearing, can almost be regarded as “pressuring”, due to its obvious significance (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“Cisco is the worldwide leader in networking for the Internet with revenues currently approximating $17 billion per year. We are a multinational corporation with more than 26,000 employees in 200 offices and 55 countries. In the U.S., we have significant operations in California, Texas, Massachusetts and North Carolina.”.

accounting for goodwill after the business combination is completed.”.

While recognising the perennial standstill of the discussions surrounding business combinations accounting, the Cisco Controller put an emphasis on the issues raised by the compulsory use of purchase method, regarded as problematic (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“These problems of the purchase method are still with us, and the implications today are much more severe than they were in 1970. In 1970, most of an acquisition price was allocated to tangible, hard assets. Today, for knowledge-based technology companies, most of the acquisition price is allocated to intangible assets - and very little allocated to hard assets. For example, since 1993, Cisco has acquired 50 companies amounting to $19 billion. Of these acquisitions, only $900 million, or 5%, is attributed to hard assets - $18 billion or 95% would be left to allocate to intangible assets or goodwill. So because over time the nature of acquisitions has changed in the New Economy, the limitations of the purchase method have become more problematic. And yet the new FASB proposal would force all acquisitions to be accounted for under the purchase method, without having solved its defects.”.

As in the case of James Barksdale, the existing purchase method, as of APB Opinion 16 and 17, and the FASB’s proposals on a revised purchase method (Financial Accounting Standards Board, 1999a) were regarded with major concerns, as pooling would no longer be an option (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“One of the defects of the purchase method is the accounting for goodwill once it is recorded as an asset on the balance sheet. The FASB proposal requires that goodwill be treated as a wasting asset, and be amortized ratably over 20 years. This model incorrectly assumes that goodwill declines in value over time, which artificially reduces net income and misrepresents economic reality. In reality, the value of goodwill is dependent upon the success of the merger, and is not a function of time.”.

Examples of the negative impact of purchased goodwill amortisation in 20 years on successful M&A deals were also provided by means of simulation (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“For example, we studied four technology mergers that occurred in 1996 and 1997, which were reported as poolings. We then recast the poolings as if they were purchases, and restated the financial statements for periods after the acquisition to show the impact of goodwill amortization. Using the Computer Sciences merger with Continuum as an illustration (…), the impact of the goodwill amortization is to reduce actual net income by an average of 18% each year. This would suggest that goodwill has declined in value. However, over this same time period, goodwill actually increased by 42%. The results of all four companies in the study, (…) demonstrate the same impact. The purchase accounting model significantly reduced actual earnings by an average of 48% for the amortization of goodwill, which presumes that goodwill is declining in value. But in reality, the goodwill has significantly increased from the date of the merger by an average of 43%. In these examples, which are typical of successful mergers, the purchase model grossly misrepresents economic reality.”,91 also stressing that the possible impact could be dramatic in the case of unsuccessful M&A deals:

“On the other hand, not all mergers are successful. In one merger we studied, the goodwill declined from $7 billion to $1.6 billion in just three years after the acquisition.”, being the views on goodwill regarded as critical to justify such possible severe impacts:

“Based on the above study, it is clear that in successful mergers, the presumption that goodwill is a wasting asset is not valid.

91 Original underline.

Goodwill increases in successful acquisitions and declines rapidly in unsuccessful acquisitions. One thing goodwill does not do is decline ratably over twenty years - the FASB model simply does not report true economic performance. Similar observations may be made of other intangibles, such as brand names, where the value for many brands have increased - not declined - over time.”.

Further considerations were also made about the proposed accounting treatment for purchased intangible assets. While warning about the need of abundant guidance and standards about how to identify and value intangibles, Dennis Powell firmly stated that (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“At risk is a loss of credibility in financial reporting.”.

Similarly to James Barksdale, FASB’s proposals were considered to do not solve the missing comparability resulting from the existing of accounting choice in business combinations accounting. Nevertheless, a different type of examples were provided (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“The FASB has stated that elimination of pooling solves a comparability issue between purchase transactions and pooling transactions. But elimination of pooling simply trades one comparability issue with a set of new comparability problems. First, mandating the purchase method creates significant comparability issues between companies who grow from internal organic development and those who grow through acquisition. For example, a company that generates significant goodwill from its internal operations will report no goodwill value while the company that acquires goodwill through a merger will report the "value" of the goodwill at the time of the acquisition. So, while both companies may have the same value of goodwill, only the company who obtained the goodwill through a merger will report any amount on its balance sheet. Secondly, elimination of pooling prevents comparability within the same company - in comparing operations before the acquisition, which do not include the activities of the acquired company, to

operations after the acquisition, which do include the activities of the acquired company. Eliminating pooling does not solve the comparability issue.”, to finally argue that it would be more important to improve purchase method, than to eliminate pooling of interests:

I believe the comparability issue would be more effectively addressed by correcting the inherent problems of the purchase method than by eliminating pooling accounting as an option.”.

To conclude, at a time when the “new economy” was a fashionable buzzword, the Cisco Controller made a warning on the economic impacts of FASB’s proposals, namely in terms of technological advances, capital formation, and jobs creation (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“In conclusion, the U.S. accounting rules for Business Combinations, which includes both the pooling and purchase methods, has for the past 50 years, generated and supported the strongest capital markets in the world. Before the FASB radically changes these accounting rules to a model that will certainly stifle technology development, impede capital formation and slow job creation in this country, the FASB should make sure the proposed new method is without question, the absolute, correct solution. In reality, the FASB’s proposed standard does not improve the accounting - it merely changes it. Worse yet, the proposed changes require companies to use a purchase model that does not work for companies in the New Economy, where most of the acquisition value cannot be attributed to hard assets, forcing companies to report an arbitrary, artificial net income number that is irrelevant and misleading.”.

As an epilogue, Cisco’s testimony ended with the following proposals (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000):

“We believe the FASB should:

100

(1) Retain the pooling of interests method of accounting. (2) Revise the purchase method to correct its deficiencies (…) (3) Engage a task force, which would include valuation experts, to develop adequate guidance on how to identify, value and account for intangible assets for New Economy companies”.

This is, besides the suggestions referred to before, a specific accounting treatment for the “new economy”, hi-tech, companies would also be very much welcomed by the business community.

Similar arguments and allegations were made in the remaining testimonies. For example, John Doerr, a partner with the venture capital firm Kleiner Perkins Caufield & Byers, serving on the on the board of directors of 15 companies, including Amazon.com, Sun Microsystems, Excite@Home, Healtheon/WebMD, Intuit, Martha Stewart Living Omnimedia, Freemarkets and Handspring, and “perhaps America’s best-known venture capitalist” (Beresford, 2001: 75), argued that many M&A deals between many of these companies would not go through without pooling (Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000):

“Without pooling, many of these mergers could not have taken place due to the negative effects of goodwill amortization on earnings. The elimination of pooling will constrain companies from engaging in business combinations that make sense.”.

While purchase method was also attacked by Doerr – similarly to other speakers (Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000):

“Technology companies concede that there are issues associated with the treatment of intangibles. Today’s purchase accounting method, however, is not the answer. It does not solve the intangible asset issues, adequately represent corporate performance, or reflect the value of a successful business combination.”,

101 any purposes of corporate lobbying or politically interferences in the business combinations accounting topic were dismissed:

“I am not here to ask Congress to make accounting rules or to ask FASB to make economic policy. Only that we work together to undertake a thoughtful, fundamental review of the treatment of intangible assets that does not derail innovation and economic growth.”, despite direct references to undergoing interferences at the SEC, trough the Garten Commission, with possible good compromising results at FASB’s work on M&A accounting:

“As many of you know, Securities and Exchange Commission Chairman Arthur Levitt has asked Jeffrey Garten, Dean of the Yale School of Management, to assemble a team of leaders from the business community, academia and the accounting profession to consider how our current business reporting framework can more effectively capture the historic changes that are taking place in our economy. I am pleased to be a member of the Garten Commission and I am committed to working with the FASB to address the intangibles issue. Efforts such as this will deliver both transitional and permanent solutions that enable a well thought out, prepared evolution to improved accounting methods and systems.”,

Accordingly, Doerr ended his testimony with a friendly and optimistic approach to the FASB (Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000):92

“Under Chairman Jenkins’ leadership, the Board has raised some important issues that require attention. I believe there is a process in place that can address these complex issues - through the Garten Commission’s review of the financial reporting model. In

92 Doerr refered to Edmund Jenkins as “my friend” (Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000).

102

the near term, these efforts will allow a well thought out model to be developed that will improve the current system. We have a lot of work to do together and we in the technology community look forward to participating in that dialogue. Only then can we make sound decisions about the future of pooling and the appropriate treatment of intangible assets without threatening our economy’s tremendous growth and the value of financial statements to investors.”,

It is therefore arguable that the business community was confident that FASB would be forced to compromise, as both the SEC and the Congress were already deeply involved in the debates regarding the business combinations accounting issue. And the scale plates appeared to be increasingly unbalanced against an underweight FASB. This suggestion is also supported by Senator Gramm’s opening statement on the “The Pooling Method of Accounting for Corporate Mergers” hearing. According to Beresford (2001: 74), its opening statement included this straightforward assertion:

“As I look at accounting, no one could ever make a claim that accounting is reality. No one could ever make a claim that accounting is based on precise theory where we reduce a very complex reality down into a balance sheet that perfectly reflects it. The goal of accounting is basically to come up with simplifications that reflect reality and, to the extent that any simplifications ever reflect a complex reality, produce a situation in which you can get a good view of what is actually happening to a firm in the marketplace from the accounting standards that were set.”.

Interestingly, despite almost entirely overruling the accounting role, and inherently dismissing the role of the accounting standard-setting body, a remark is made about the importance of preserving FASB’s independence (Beresford, 2001: 75):

“there is nobody more committed to the independent setting of accounting standards than I am. But I continually question the independence of the setting of these standards. I think the Securities and Exchange Commission has too much to do with the setting of these standards.”.

103

Therefore, doubts were cast about the efficacy of FASB as a standard setter, perhaps due to SEC interferences – which is also often effectively a target of corporate lobbying, particularly whenever lobbying on the FASB does not produce results.

Also according to Beresford (2001: 74), the opening statement of Senator Gramm included not only his own opinion on accounting issues, as he has also suggested “not mandating goodwill amortisation but instead requiring a periodic review for impairment”. This is a very significant suggestion, as later FASB would indeed replace the proposed 20 year ceiling amortisation period by impairment tests.

It is also important to refer that the pressures on the Congressional hearing were not made only from IT and financial industries.93 The representative from the National Association of Manufacturers (NAM), Kimberly J. Pinter, has also disagreed with FASB’s proposals (Testimony of Kimberly J. Pinter, Director, Corporate Finance and Tax National Association of Manufacturers On behalf of the National Association of Manufacturers before the Committee on Banking, Housing, and Urban Affairs United States Senate On Business Combinations and Intangible Assets, 2000):

“The centerpiece of this project is the proposed disallowance of the pooling-of-interests method of accounting. The NAM finds this proposal objectionable based on a number of different factors.”,

93 e.g. Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000; Testimony of Alain J. Hanover, CEO, InCert Software Corporation, to the Senate Committee on Banking, Housing and Urban Development on behalf of The Massachusetts Software and Internet Council on Pooling of Interest Methods of Accounting, 2000; and Prepared Testimony of Mr. Harvey Golub Chairman and CEO American Express, 2000; Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000; for IT and financial industry, respectively.

104 referring similar objections to other speakers in the hearing, and also expressing serious concerns about possible negative consequences on M&A activity:

“I have personally discussed the proposed elimination of pooling with many of our member companies, and I have been truly surprised by the number of times I’ve heard that this merger or that merger would not have happened had it not been for the applicability of pooling. And I’ve heard these comments across the board from all kinds of manufacturers. Even those that don’t use pooling are very concerned about its possible unavailability for future transactions.”, that could consequently be reflected negatively in the manufacturing industry and in the economy overall:

“Manufacturing is the largest contributor to economic growth, and the recent surge in M&A activity has coincided with a surge in productivity growth. By mentioning these facts, I don’t mean to suggest that pooling should be retained because it somehow "encourages" business combinations; rather, it appears that the existence of only the purchase method to account for a diverse array of transactions would discourage such activity - and that result could well have a negative effect on the economy.”.

While “supporting” FASB’s independence, Kimberly Pinter and the NAM expressed also concerns about FASB’s due diligence (Testimony of Kimberly J. Pinter, Director, Corporate Finance and Tax National Association of Manufacturers On behalf of the National Association of Manufacturers before the Committee on Banking, Housing, and Urban Affairs United States Senate On Business Combinations and Intangible Assets, 2000):

“Finally, the NAM is concerned that the FASB is not hearing from all parties who may be critical of the project. Too often we have found that companies are very reluctant to too visibly criticize the merits of a FASB proposal due to concern that such activity might invite increased SEC scrutiny. Regardless of whether such concerns are founded, as they say, perception is reality, and it does have a chilling effect on full participation in

105

the process. That said, the NAM appreciates the FASB's extensive efforts to thoroughly evaluate these issues with significant outside input and participation. We do fully support the FASB's independence and private-sector setting of accounting standards.”.

Representatives from the US Chamber of Commerce and from healthcare industry completed the hearing, and expressed concerns similar to the remaining speakers. For example, the US Chamber of Commerce Chief Economist recommended FASB to drop its proposals, warning about possible constraints in the economic growth (Prepared Testimony of Mr. Martin Regalia Chief Economist U.S. Chamber of Commerce, 2000):

“The technology and financial services sectors have played a crucial role in our current position as a world leader. One reason for the success of these sectors has been their ability to grow. We do not believe that this opportunity for economic growth should be curtailed because of staffing constraints or the desire to adjust our standards to conform to our international competitors, especially when there is no consensus abroad. We should not change our standards without more compelling reasons. The U.S. Chamber urges FASB to reconsider its position and withdraw its exposure draft on this issue until Chairman Levitt's group has reported and all concerned parties have had the opportunity to digest its findings. The issues at stake are of great concern to our members and to all who want to encourage the continued economic growth we are currently enjoying.”.

Similarly, the Medtronic CFO urged FASB to change its proposals on M&A accounting, while expressing concerns about the possible generation of “uncertainty” and “volatility” on the markets, as a consequence of FASB’s confusing new rules (Prepared Testimony of Mr. Robert Ryan Chief Financial Officer Medtronic, 2000):

“We urge the FASB to defer decisions on the accounting for business combinations until the whole issue of reporting financial performance is addressed. We think that the FASB's position relative to business combinations may change. We believe that the current proposal will create confusion in the marketplace, and

106

uncertainty creates volatility. We have been told that investment analysts will train the market to ignore goodwill amortization. I believe we should be providing the market with meaningful and useful information and not something that will create confusion and be ignored.”.

As mentioned before, FASB defence was solely in charge of its head, Edmund Jenkins (Beresford, 2001). To be more precise, he was accompanied by Kim Petrone, one of the business combinations accounting project managers, but without any speaking role (Hopkins et al., 2000). The FASB’s chairman defended the ED proposals (Financial Accounting Standards Board, 1999a), which have been examined before in this chapter, emphasising the need to improve the financial reporting quality of M&A accounting. Nevertheless, it also revealed ample openness for a further discussion of the initial FASB proposals (Hopkins et al., 2000):

“The Board has not yet made any final decisions with respect to the requirements contained in the proposal. The next significant stage in the Board's due process will be to begin its redeliberations of all of the issues contained in the proposal, including the Board's proposed decision to eliminate the pooling method. That process will include numerous public meetings held over the next several months. At those meetings, the Board will carefully consider the comment letters, public hearing testimony, what we learned from this hearing, and all other relevant information received from constituents - including information obtained from the members of our Task Force on Business Combinations and the members of our Financial Accounting Standards Advisory Council. No final decisions will be made until that process is completed.”.

An indication that FASB could concede to the growing pressures was also given (Hopkins et al., 2000):

“In closing, I want to be clear that the FASB understands and supports the oversight role of this Committee. We will carefully consider what we learn from this hearing. Let me assure you, Mr. Chairman and members of the Committee that our open due process and our independent and objective decision making will be

107

carefully and fully carried out. To do otherwise would jeopardize the very foundation upon which the FASB was created, and for which it has proven invaluable to the US capital markets and to investors - the consumers of financial information.”.

History tends to repeat itself, but lessons can be drawn, and certainly FASB did not want to face the same fate as AICPA’s APB.

A second hearing entitled “Accounting for Business Combinations: Should Pooling Be Eliminated?” was carried on 4 May 2000 with Michael Oxley, Chairman of the Subcommittee on Finance and Hazardous Materials of the Committee on Commerce. There were again two panels, but this time the second panel included not only Edmund Jenkins, but again Dennis Powell, and, most notably, Peter Bible, Chief Accounting Officer of General Motors, who was in favour of eliminating pooling (Beresford, 2001).

Nevertheless, concerns with pooling elimination and possible economic impacts from FASB’s proposals continued. For example, while highlighting the importance of FASB’s independence, the hearing convenor stated that (Beresford, 2001: 77):

“I would like to begin by reaffirming my belief that FASB, as an independent private sector entity, is best suited to set accounting standards. Few would want politicians without accounting expertise making highly technical accounting decisions.”, however, Michael Oxley would continue with the same tone given by Senator Gramm in the previous hearing:

“From time to time, however, FASB considers an issue which has broad public policy implications, best brought to light through Congressional bearings. In such an instance Congress has a responsibility to foster open dialogue on the issue. Though I do not begin to have a solution to this debate I do urge those central to this debate to consider all options. Perhaps we need to further examine, the changing nature of assets which is driving our

108

economy. We need to consider whether eliminating pooling accounting has a negative economic impact that could diminish the competitiveness of US businesses. Finally, we must evaluate whether the proposed changes will actually provide investors with more useful information about a company.”.

Representative John Dingell, participating in the first panel of this second hearing, urged to the achievement of a compromise between the contending parties (Beresford, 2001: 78):

“I urge the Board to proceed cautiously and carefully and, weighing the costs and the benefits, to act to achieve incremental good. I would also urge my friends in the hightech industry to work with FASB to develop a compromise that eliminates the current biases and distortions.”, while, conversely, Representative Bob Goodlatte continued to express serious concerns - that eliminating pooling accounting would strangle growth of the new economy - by stating:

“I am hopeful that the FASB will step back, take a deep breath, and see the forest that is the New Economy, rather than the trees that are individual accounting standards. I look forward to working with you [Representative Oxley] and others who are concerned that our system of accounting standards should move along with the rest of the economy into the new century”.

Furthermore, Representative Dooley strongly supported Congressional oversight of the business combinations project, as he directly targeted not only the FASB, but also the FAF (Beresford, 2001: 78):

“In the end, Congressional oversight on contentious issues does not weaken the process, it strengthens it. What weakens the process and its product is when FASB stubbornly ignores the concerns of its constituents; and when the Financial Accounting Foundation, which oversees, funds, and appoints the members of FASB, characterizes Congressional interest and concern about a FASB

109

project as "Explicit or implicit threats..." as they have in a May 1, open letter.”.

Regardless all these pressures, Edmund Jenkins continued defending FASB views, particularly regarding pooling elimination (Beresford, 2001: 79):

“The Proposed Standard’s requirement to eliminate the pooling method will benefit investors, creditors, and other financial statement users by providing more information and more relevant information about all mergers and acquisitions. The Proposed Standard’s provisions also will benefit those consumers by improving the comparability of financial reporting, thereby making it possible to more easily contrast companies that participate in business combinations.”.

But FASB’s views continued in peril. Along with the hi-tech industry, there was the financial sector, particularly the investment banks which were participating in the merger mania. As main financial advisors and financers of M&A deals, the banking industry was not certainly interested in losing business, simply because of a change in the accounting regulation, and therefore continued developing an intense political lobbying (Zeff, 2002).

The banking industry was not only fuelling the M&A wave by financing deals, as it was also undergoing a consolidation process itself. Banks also favoured pooling of interests, particularly in large dollar-value deals (Schroeder, 1999: 7). Indeed, according to Ayers et al. (2000: 8), “acquisitions in the financial services industry account for approximately one-third of all pooling acquisitions in number in value”. Historically, banks could afford to pay higher premiums based on book value for M&A deals to be accounted by pooling of interests (Schroeder, 1999: 12). By being necessarily forced to use purchase method, the banking industry would not be able to continue to offer high premiums bids, and therefore M&A deals among banks would be reduced. It was therefore inevitable that the financial industry would strongly oppose to pooling discontinuation.

110

As a natural outcome, the respondents to the FASB’s early Invitation to Comment from the banking industry strongly opposed to the pooling elimination proposal (Ayers et al., 2000; Financial Accounting Standards Board, 1999b). As much as 70% of the banks defended the continuation of both methods in order to respect the broad range of business combinations types (Financial Accounting Standards Board, 1999b).94

Like AICPA in 1970, FASB had also to compromise. On 14 February 2001, the FASB issued a Revised Exposure Draft, Business Combinations and Intangible Assets - Accounting for Goodwill, in which keeping pooling elimination was proposed (Financial Accounting Standards Board, 2001b). However, purchased goodwill and indefinite lived intangible assets were no longer to be amortised but to be tested annually for impairment. This was an amusing solution for the companies that lost the pooling option and had to recognise large amounts of purchased goodwill, as the goodwill amortisation charges were eliminated. It was also considered a more consistent decision, given the arbitrary nature of the initial proposal of a 20-year ceiling for goodwill amortisation.

Despite the compromise, the FASB has received 211 response letters on the 2001 Exposure Draft (Financial Accounting Standards Board, 2001a). Over 20% of the comment letters, 42 responses, were from the financial industry (Lewis et

94 There is not comprehensive data available about pooling usage in hi-tech and financial industries. Nevertheless, as examined, anecdotal evidence supports the suggestion that pooling of interests was largely favoured in these industries, and therefore one can consider the likelihood of a considerable level of pooling usage (e.g. Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000; Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000). Anecdotal evidence also suggests that pooling was desirable in other industries, such as manufacturing, or healthcare (e.g. Prepared Testimony of Mr. Robert Ryan Chief Financial Officer Medtronic, 2000; Testimony of Kimberly J. Pinter, Director, Corporate Finance and Tax National Association of Manufacturers On behalf of the National Association of Manufacturers before the Committee on Banking, Housing, and Urban Affairs United States Senate On Business Combinations and Intangible Assets, 2000).

111 al., 2001). Therefore the discussion about the new business combinations accounting seemed not over yet.95

As the final issues were finally addressed, by May 2001, the board could finally vote the new standards, SFAS 141 and SFAS 142, by the end of June. Unlike APB Opinions 16 and 17, FASB’s statements were adopted by the unanimous vote of the six members of the board. It was the result of 5 years of work, from 1996 through June 2001, and discussions, which included over than 60 public meetings and the discussion and analysis of more than 500 comment letters (Financial Accounting Standards Board, 2001d).

Finally, pooling of interests was eliminated in the USA, with other accounting- -setter boards around the world following G4+1 recommendations and FASB’s new GAAP.96 In December 2007, FASB issued a revised version of SFAS 141 which replaced the original 2001 version. The new standard, SFAS 141(R), was issued together with IASB’s IFRS 3, Business Combinations, which was also subject to revision (International Accounting Standards Board, 2008).97 These two standards were the result of a joint project by the FASB and the IASB, “to improve financial reporting about business combinations and to promote the international convergence of accounting standards.” (Summary, Financial Accounting Standards Board, 2007a). Unlike the troubled APB Opinions 16 and 17, the new set of FASB’s business combinations and goodwill accounting standards seemed to be far more successful, despite several issues regarding its

95 From this point, in this thesis “new business combinations accounting standards” refer to both SFAS 141 and SFAS 142, despite being SFAS 141 the only one entitled as “Business Combinations”. 96 The “Group of 4 plus 1” was an avant-garde accounting mastermind group intended to promote international GAAP convergence. Founded as an international group of standard setters, it was composed by the Australian Accounting Standards Board (AASB), the New Zealand Financial Reporting Standards Board (FRSB), the UK Accounting Standards Board (ASB), the Canadian Accounting Standards Board (AcSB), and the FASB (G4+1 (Organization) & International Accounting Standards Committee, 1998). The IASC has participated in G4+1 works as well, but with a different status, as it was only an observer. 97 IFRS 3 (R) was issued in January 2008.

112 implementation. Some sensitive topics included the application of purchased goodwill impairment tests, or the intangible assets allocation. However, the further guidance provided in order to ease and clarify its real application cannot be compared with the AICPA’s regulatory myriad in the 1970’s. Perhaps the best proof of the appropriateness of the new standards is that the 2007 FASB’s revision retained the fundamental requirements of SFAS 141, although it changed the name of the purchase method to acquisition method. SFAS 141 (R) has not only replaced SFAS 141, but has also amended SFAS 142 in several matters. It also provided further guidance in order to improve the application of the purchase method, and filled the gap on the accounting for bargain M&A, i.e., transactions that result in negative goodwill.98

Despite the apparent success of the implementation of FASB’s new business combinations standards, particularly when compared with the previous set of standards issued by the AICPA, the question of whether such changes had any significant impacts on M&A activity remains unanswered. Indeed, formal implementation success does not necessarily mean success in limiting impacts on the economic activity, and therefore one can argue that the economic effects issue has enough merits to justify its study, as aimed by this thesis.99

2.7 Conclusions

The quest for accounting principles, which could integrate a comprehensive theoretical set, resulted in the establishment of a conceptual framework driven primarily by pragmatism, prompting discussions among the accounting community, while stimulating pressures from interest parties.

98 As SFAS 141 revision does not result on substantial effects regarding the purposes of the present research, and therefore the new standard is not to be subject to any further examination. 99 Implementation success considered in terms of accounting choice elimination, as some issues, particularly the discretionary use of purchased goodwill impairment methodology, continues to raise doubts among researchers (e.g. Ramanna, 2008).

113

It appears that financial reporting is far too important simply to let it be managed by the accounting profession or by independent experts. This importance is not exclusive to managerial executives, who are concerned with their performance measurement. Politicians also want to see their constituents pleased, and are therefore concerned about the way the metrics of the economy are constructed – among others, economic performance can also be considered as a performance indicator for politicians itself.

Consequently, companies have clear incentives for lobbying, as long as the possible outcome outweighs the costs of doing so. The FASB’s open-way conduction of its standard-setting policy also encourages lobbying. Therefore, as examined in this chapter, companies do lobby FASB whenever their interests are in peril, and as long as they can achieve net benefits from such actions. If they are unsuccessful with the FASB, they may try to influence the SEC, and even the Congress of the USA if necessary, in order to do as much as possible to persuade the FASB to concede to their interests.

The business combinations accounting issue illustrates perfectly this dichotomy, which often occurs between financial reporting standard-setters and preparers. For decades, almost since its inception, that pooling of interests was regarded with suspicion by the accounting profession, while loved by the business community, so much so, that its lobbying was successful to avoid pooling elimination in a first real attempt, in the 1970’s. On a second attempt, however, pooling elimination could not be avoided, despite some fierce lobbying on the FASB, on the SEC, and on the Congress of the USA. Nevertheless, in order to achieve such a victory, not only has FASB been subject to every sort of pressures, but it also had to compromise inclusively, by abandoning its proposal of goodwill amortisation. Whether FASB’s concession was enough to mitigate the negative effects of pooling elimination, as referred to by the business community, is the question to be examined in this thesis.

114

Chapter 3 M&A Activity and M&A Accounting

3.1 Introduction

This chapter presents an overview of the evolution of M&A activity and of the recent developments on business combinations accounting. It starts with a note on M&A terminology, followed by a brief presentation of some key definitions.

As mentioned earlier, it was not possible for researchers to explain why M&A tends to occur in waves. Nevertheless, the historical pattern of M&A is well documented, and therefore M&A waves are described in this chapter according to existing data and literature.

While the earlier evolution of the accounting for business combinations in the USA was examined in Chapter 2, in this chapter an analysis of the changes in the regulation which occurred in 2001 is made.

Finally, following an examination of some issues related to accounting neutrality, an early assessment of the possible economic consequences that may arise from the changes in M&A accounting is made, together with its possible impacts on financial reporting. This topic constitutes the core of the present research, and therefore it will continue to be examined throughout this thesis. Additionally, by continuing the discussion of why it had not been possible so far to construct a comprehensive conceptual theory for accounting and for M&A activity, this chapter also contributes to justifying the need for a greater

115 understanding of these phenomena. Such missing acquaintanceships also help to substantiate the need for the development of specific models for this thesis.

3.2 Terminology and definitions

Terminology is prolific and diverse in M&A, and therefore can be confusing in terms of precise meanings. In fact, M&A concepts are often used arbitrarily, even imprecisely, not only in professional contexts, but also in more formal contexts, such as in academic ones (vid. e.g. Weston et al., 1998). This is natural, as M&A jargon used by finance professionals, and in other informal contexts like the media, arbitrarily merges with more formal nomenclature used in legal, and in other normative contexts, such as tax, or accounting. Authors such as Weston et al. (1998), or Gaughan (1999; 2002), identify some discrepancies among M&A types of deals definitions. Since M&A terminology can be ambiguous, it is therefore advisable to define some key concepts more accurately, even if such definition relies on a criterion which is arbitrary itself. In order to minimise such subjectivity, different definitions used in M&A literature were crossed as follows.

M&A refer primarily to mergers and acquisitions. Merger is understood to be an operation that results in a single entity from the combination of two or more entities. The resulting single company is often one of the previous entities that survives at the expense of the disappearance of the remaining entities.100 The surviving company can be regarded as an acquirer, since it will absorb all the remaining companies involved in the merger. In case the acquirer assumes the assets and liabilities of the merged companies the deal can be formally defined as a statutory merger, while if a target entity becomes a subsidiary of the acquirer it is called a subsidiary merger (Gaughan, 2002: 7). When the merger results in the formation of an entirely new single company, the deal can be referred to as a consolidation (Gaughan, 2002).101

100 Merger (statutory, subsidiary): Entity A + Entity B + … + n entities = Entity A. 101 Consolidation: Entity A + Entity B + … + n entities = Entity Z ≠ A, B, … , n entities.

116

In an acquisition, one entity acquires the assets or the shares of another, losing the shareholders of the target entity the ownership. The resemblance of acquisitions to mergers is obvious, leading many to argue that, substantially, consolidations are the true form of mergers.102

Mergers usually result from a previous agreement between the companies involved. This is not necessarily the case for acquisitions, which can be classified as friendly or hostile. Friendly acquisitions are negotiated with the management of the target entity to meet certain legal and technical requirements, while hostile bids involve a direct offer to the target’s shareholders to sell their shares at a specified price without the target’s board of directors’ approval (Weston et al., 2004). Hostile acquisitions are characterised as unsolicited bids that usually involve vigorous public rejection by the target’s management (see e.g. Andrade et al., 2001; G. F. Davis & Stout, 1992).

M&A is also referred to as a takeover, although ‘takeover’ is a generic and vague term (see e.g. Gaughan, 2002; Ross et al., 1996). According to Gaughan (2002: 7), sometimes it refers only to hostile transactions, while at other times refer to both friendly and hostile deals. With a different understanding, Sudarsanam (2003: 3), argues that a takeover resembles an acquisition that also implies a “acquirer much larger than the acquired”. Takeover is also frequently used interchangeably with tender offer (Weston et al., 1998: 4). Indeed, a tender offer is a method of making a takeover via a direct offer to target entity shareholders (Weston, 2001), and therefore the confusion of terms is natural.

Although M&A is essentially mergers and acquisitions that result in a change of control, it is no less true that M&A goes much further beyond this. Joint ventures and strategic alliances; restructuring activities such as divestitures, carve-outs, spin-offs, and tracking stocks; changes in ownership structure through share repurchases, leveraged buyouts, dual-class recapitalisations, and

102 i.e. statutory mergers and subsidiary mergers can be regarded as mere acquisitions. Besides consolidations, mergers of equals are also considered to be genuine mergers.

117 leveraged recapitalisations are some examples of types of operations that can be broadly related to M&A.103

Grinblatt & Titman (1998: 669) point out that there are probably as many types of mergers as many different bidders and targets. Nevertheless, they can be epitomised as horizontal, vertical, or conglomerate. Horizontal mergers involve firms operating in similar businesses, such as competitor firms (e.g. Bank of America and FleetBoston Financial Corp.). Vertical mergers occur in different stages of production chain (e.g. AOL and Time Warner). Finally, conglomerate mergers involve firms from unrelated business activities (e.g. General Electric and Abbott Laboratories).

Finally, in the present research, the terms M&A and business combinations are used interchangeably, as synonyms. M&A is used in finance, while business combinations have been adopted by accounting standard-settings boards. Although M&A and business combinations may refer to several different types of operations, they both refer essentially to mergers and acquisitions. This is also the view adopted in this research.

3.3 M&A activity pattern and M&A waves

This section of the thesis examines the pattern of M&A activity and also the issue of the M&A waves occurrence, a peculiar M&A behaviour that continues to generate discussions among researchers. This examination is relevant to the present research as a fair understanding of the pattern of M&A occurrence is important to assess whether the changes in accounting regulation resulted in a significant impact on the level of M&A activity, as to be tested afterwards in this thesis. Therefore this section starts with an analysis of the historical pattern of the M&A waves, followed by the study of some of the determinants in M&A activity, which have been selected in the scope of this research’ purposes.

103 For definitions of different types of deals related to M&A vide chapter 6.

118

Accompanying the economic and entrepreneurial environments, the evolution of M&A activity tends to incur in cycles, just like the economic ones. Sudden rises and falls, such as M&A deal peaks quickly followed by crashes, are not rare in the M&A activity, resembling the stock exchange markets volatility. As in the market bubbles, sometimes the M&A activity is so intense that it leads to a so called ‘merger wave’. This pattern of behaviour is not new. Although there are different sources and sets of M&A data, they all confirm the existence of several peaks of activity during more than a century of history in M&A. It is therefore accepted that M&A activity through time seems to come as in waves. The casual observation of historical M&A data indeed suggests this wave-like pattern, as shown below.

Fig. 3.1 Number of mergers completed in the USA, 1895-1990 Opere citato Stearns & Allan (1996: 700). Data sources: Thorp (1941); R. L. Nelson (1959); U.S. Department of Commerce (1975); Mergers and Acquisitions (1981, 1991) (op. cit. Stearns & Allan, 1996: 700).

119

Most authors, such as Weston (2001), or Gaughan (2002), identified five M&A waves in the USA since the 19th century, being the first four movements represented in Fig. 3.1.104 The M&A activity and waves are well documented. For example, the two first waves were studied extensively, either in conjunction or separately, by Thorp (1941), Stigler (1950), Markham (1955), R. L. Nelson (1959), Weston (1961), Eis (1969), or more recently by Lamoreaux (1985), and Leeth & Borg (2000); while the third wave has been studied by R. L. Nelson (1966), Reid (1968), Lintner (1971), Lynch (1971), Markham (1973), Steiner (1975), Scherer & Ross (1990), Hubbard & Palia (1999), among others. The most recent waves, the fourth and the fifth, solo or together with previous waves, were reviewed by Ravenscraft (1987), Golbe & White (1988, 1993), Blair & Schary (1993), or Mitchell & Mulherin (1996); and Mueller (1997), Black (2000a, 2000b), Andrade et al. (2001), Holmstrom & Kaplan (2001), Jovanovic & Rousseau (2002), or Harford (2005); respectively.

Although the wave-pattern was similar in each merger wave, including prominent peaks of activity, each of the waves was driven differently according to the entrepreneurial and economic environment existing at the time. The M&A movements are not casual, as specific changing factors and development forces instigate and give them particular characteristics. Indeed, all the waves were driven by different reasons, and had different characteristics. For instance, the type of deals and the most active industries have varied at different moments in time. A brief review of the history of M&A waves in the USA follows, based on Weston (2001: 7-8).

The first merger movement occurred at the turn of the century, from 1893 to 1904, and was associated with the completion of the transcontinental railway system. It created the first common market as it favoured the creation of

104 The analysis of the fifth wave, occurred in the late 20th century, is made immediately after the examination of the previous waves, and is accompanied by a figure that allows its comparison with the fourth wave. The gigantic scale of the fifth wave, particularly when compared with the earlier waves, makes it inconvenient to exhibit all five waves in a single graph, as visual observation would be distorted, even if plotted in a logarithmic scale.

120 monopolies. Major horizontal mergers occurred in steel, oil, and basic manufacturing industries. According to Fligstein (1990), between 1895 and 1904, 78% of the mergers were horizontal, 10% horizontal and vertical, being the remainder solo vertical M&A.

The second merger wave occurred in the 1920s. According to Gaughan (2002), between 1916 and 1929. This merger wave was prompted by the development of the radio and the automobile, which made national advertising possible, allowed an increase in wider geographic sales, and it also helped to organise distribution. It was characterised by an increase in vertical mergers, which enabled manufacturers to control distribution channels in a more effective way. This led to the creation of large conglomerates, in an economy ruled by oligopolies. The second M&A wave ceased dramatically with the 1929 crash.

“Conglomerate mergers” is the best label for the third merger wave, which occurred in the late 1960s. As the post-war effort was reducing considerably, the conglomerate mergers represented, in part, a response to the slowdown in defence expenditures. As Weston (2001: 8) refers, “in every sample of conglomerates, at least one-half of the companies were aerospace or natural resource – depleting companies (oil, forest)”. A new euphoria took place in the markets, as it was believed that a good manager, supported by the planning offered by state of the art literature, could manage anything successfully. It was also a period willing to diversification. For example, diversification was used by food industries in order to avoid their growth being too much linked down to population growth. This period, where diversification was fashionable, led many companies to overlook their core businesses, resulting in major decreases in corporate price-earnings ratios (PER) by the time the wave was vanished.105

105 From the analysis of a sample of conglomerate mergers, Weston & Mansinghka (1971) found that deals were not made to enhance profits, but to increase earning per share (EPS) and shares prices, through a favourable PER. This finding would be later confirmed by Conn (1973). For more insights on the topic see also Mead (1969), Weston (2001) and Weston et al. (2004).

121

A fourth merger wave took place in the 1980s. It brought a certain dismantling of the diversification wave of the 1960s. In the 1980s it was the era of tender offers and hostile takeovers, which were enhanced by financial innovations and availability of high-risk financing. High leverage, junk bonds, arbitrage and speculation, and a higher involvement of investment banks, resulted in any firm to be possibly vulnerable to a takeover bid, as long as it was not performing up to its potential (Weston, 2001: 8).106

The M&A wave which occurred in the last decade of the past millennium was a remarkable one. From 1993 the number of transactions and the M&A value increased in a steady way. By the end of the century, an unprecedented pace, fed by a favourable economic environment and euphoria in the markets, resulted in colossal megamergers that broke all previous milestones in terms of deals size. The overall M&A activity figures, both in number of operations and dollar value, were also all-time high. The magnitude of the most recent wave is shown below in Fig. 3.2.

106 Risk arbitrage with M&A is common and consists in making short-term gains from the relationship between the takeover bid price and the relative prices of the bidder’s and target’s shares (Weston, 2001: 10):

“For example, bidder B selling at $100 may offer $60 for target T, now selling at $40 (a 50 percent premium). After the offer is announced, the arbitrage firm (A) may short B and go long in T. The position of the hedge depends on price levels after the announcement. Suppose B goes to $90 and T to $55. If the arbitrage firm (A) shorts B and goes long on T, the outcome depends on a number of alternatives. If the tender succeeds at $60, the value of B may not change or may fall further, but the value of T will rise to $60, resulting in a profit of at least $5 per share of T for A. If the tender fails, T may fall in price but not much if other bids are made for T; the price of B may fall because it has “wasted” its search and bidding costs to acquire T. Thus, A may gain whether or not the bid succeeds. If the competition of other bidders causes B to raise its offer further, A will gain even more, because T will rise more and B will fall.”.

122

Jan 2000 = 100

125 500 100 400

75 300

200 50

100 25

0 0

-7 -1 -7 1 7 81-1 82-7 84-1 85-7 87-1 88-7 90-1 91-7 93-1 94 96 97 99- 00- 02-1 03-7 05-1

$ Value (left scale) No. of Deals (right scale)

Fig. 3.2 Monthly M&A completed deals and dollar values, 1981-2005 Data source: SDC Platinum (Thomson Financial, 2006).107

The fifth wave was the age of strategic megamergers, with an important contribution from cross-border operations. It was also focused on some particular industries that benefited cumulatively from deregulation, globalisation, and technological revolution, which enhanced changes in industry organisation. As Weston (2001: 8) notes, more than half of the M&A activity in a given year “has been accounted for by five or six industries”.

Like Black (2000a, 2000b), or Gaughan (2002), supports here the idea that it was not just one more wave based mainly in the activity in the USA, or simply a fifth merger wave. It was much more, because it involved the global economy and their participants. Therefore, it seems fair enough to label it as a truly international M&A wave.

M&A waves are not exclusively from the USA, the same way the international dimension was not exclusively from the fifth wave. For instance, in the UK,

107 Note: SDC’s data tracking coverage in the 1980’s is somewhat poor.

123

Sudarsanam (1995) documented, between 1964 and 1992, three peaks on the M&A activity. Similarly, Hughes (1993) found that merger activity peaked in the period 1950-1990. In France, Derhy (1999) identified two M&A waves between 1959 and 1992: the first between 1966 and 1972 and the second amid 1986 and 1992.

The wave-pattern behaviour is not exclusively from M&A activity either. In fact, as Rau & Stouraitis (2006) note, there are corporate events of all types that tend to occur as waves. Just like mergers, initial public offerings, seasoned equity offerings, debt offerings, and stock repurchases, constitute some examples of wave-pattern (see e.g. Bayless & Chaplinsky, 1996; Choe et al., 1993; Lowry & Schwert, 2002; Lucas & McDonald, 1990; Maksimovic & Philips, 2001). However, literature still lacks a consensus on why waves occur, despite extensive empirical and theoretical research (Rau & Stouraitis, 2006).

Indeed, why M&A occurs in waves continues to be a phenomenon not fully understood (Rau & Stouraitis, 2006; Ribeiro & Crowther, 2007b). Despite all efforts made, previous researchers have been unable to reach a consensus about the comprehensive theoretical framework that underlies M&A activity wave pattern.108 In fact, despite the existence of a wide empirical and theoretical

108 The finding that M&A occurs in waves seems to be indisputable. Indeed, almost all authors admitted it, regardless referring to aggregate or individual-industry activity (e.g. Andrade et al., 2001; Andrade & Stafford, 2004; Auster & Sirower, 2002; Banerjee & Eckard, 1998, 2001; Barkoulas et al., 2001; Bittlingmayer, 1985; Black, 2000a, 2000b; Blair, 1989; Blair & Schary, 1993; Brealey & Myers, 1984, 1996; Brealey et al., 2006; Cabral, 2000, 2005; Calomiris & Karceski, 2000; Chowdhury, 1993; Clark et al., 1988; Derhy, 1999; Dong et al., 2006; Dymski, 2002; Eis, 1969; Fauli-Oller, 2000; Gaughan, 1996, 1999, 2002; Golbe & White, 1988, 1993; Gort, 1969; Harford, 2005; Holmstrom & Kaplan, 2001; Hubbard & Palia, 1999; Jovanovic & Rousseau, 2002; Lamoreaux, 1985; Leeth & Borg, 2000; Linn & Zhu, 1997; Lubatkin, 1983; Matsusaka, 1993; Matsushima, 2001; Mitchell & Mulherin, 1996; Mueller, 1989; R. L. Nelson, 1959; Rau & Stouraitis, 2006; Ravenscraft, 1987; Reid, 1968; Resende, 1999; Rhodes-Kropf et al., 2005; Rhodes-Kropf & Viswanathan, 2004; Scherer & Ross, 1990; Shleifer & Vishny, 1990, 1991; Sirower, 1997; Stearns & Allan, 1996; Steiner, 1975; Stigler, 1950; Sudarsanam, 1995, 2003; Town, 1992; Toxvaerd, 2008; van Wegberg, 1994; Weston, 2001; Weston et al., 1990; Weston et al., 1998; Weston et al., 2004). Only Shughart II & Tollison (1984) were unable to

124 research, the literature still lacks a convincing broad theory, presenting many partial explanations instead.109

The purpose to produce a comprehensive theory that could suitably explain the M&A pattern and its evolution configures a difficult task, due to the complex specifications and combinations that interact with M&A activity. Nevertheless, there are some indicators that, when isolated, contribute to the understanding of the M&A pattern of activity. According to Weston et al. (1998), Weston in 1953 and Markham in 1955 were the first to find, using regression analysis, that M&A activity in the USA was significantly statistically correlated to stock prices.110 Afterwards, Gort (1969) formulated an “economic disturbance” theory to explain the relationship between merger activity and stock prices. His theory was based in the assumption that M&A activity was more likely to increase in periods of rising stock prices, due to expectational differences between shareholders and non-shareholders. However, this assumption requires admitting a significant market inefficiency between potential merger participants, during periods of high stock prices, which is unlikely to happen.

Other authors studied the phenomenon and tested several hypotheses, using different financial variables, in an attempt to explain the evolution of M&A activity.111 As examples of variables used in models: Tobin Q-ratio (Jovanovic & Rousseau, 2002); or stock prices and bond yields (Melicher et al., 1983). On a broader approach, Becketti (1986) studied the connections between M&A activity and macroeconomics. Other authors focused on the possible relationship

find the existence of merger waves, generating afterwards replies from Golbe & White (1988) and Town (1992). 109 For example, for more than twenty years that Brealey et al. (e.g. 1984; 1996; 2006), quoting Segall (1968) about the inexistence of a comprehensive explanatory hypothesis for the M&A wave of the 1960’s, continue to select the occurrence of M&A waves as one of the ten most relevant currently unsolved problems in finance. 110 Nonetheless, Weston et al. (1990) acknowledges mixed evidence on whether changes in the number of mergers are caused by stock price changes. 111 For example: R. L. Nelson (1959); Steiner (1975); Beckenstein (1979); Chung & Weston (1982); Melicher et al. (1983); Guerard (1985); Becketti (1986); Golbe & White (1988).

125 with the business cycle (e.g. R. L. Nelson, 1966; van Wegberg, 1994). With a different perspective, authors such as Stigler (1950) or Bittlingmayer (1985), have also studied the effects of deregulation, namely from changes in competition policy.

The majority of the studies using USA data concluded that M&A are in general both positively correlated with the stock market prices and the growth rate of Gross Domestic Product (GDP). Some of those studies also found correlation with stock prices in the long-term (samples up to one hundred years). Other studies using UK M&A data (samples up to thirty years) corroborated the results obtained in the USA.112

In terms of main categories of studies on M&A waves, it is possible to identify research on the timing (e.g. Barkoulas et al., 2001; Beckenstein, 1979; Golbe & White, 1988, 1993; Shughart II & Tollison, 1984; Steiner, 1975; Town, 1992), characteristics and motives (e.g. Markham, 1955; R. L. Nelson, 1959, 1966; Stigler, 1950; Weston, 1961), and determinants (Guerard, 1985; Melicher et al., 1983).113

The European Commission (2000) has also attempted to explain the evolution of M&A activity in the EU countries in the nineties. To achieve this, three factors were tested: economy size, the number of listed companies, and market capitalisation. However, the results obtained did not support any valid conclusions. Furthermore, the European Commission tested the variations in real long-term interest rates, price-earning ratios and share prices index. Yet again, the regression analyses did not provide results that could provide any significant relationships.

Like in previous studies focused on the long term, strong evidence was found to support explanations concerning the evolution of M&A activity over time. However, in shorter periods it was not possible to explain this type of evolution,

112 Vid. Sudarsanam (1995, 2003) and Weston et al. (1990, 1998). 113 These classifications are not meant to be rigid, as some studies have mixed characteristics.

126 either using capital market or macroeconomic factors. As the European Commission (2000) concluded, probably, short run variations in M&A activity for the EU result of “complex combinations” sector and country specific influences related.

Other strands of research included the study of motivations in M&A. The search for synergies, scale economies, or tax benefits, can also justify the occurrence of M&A.114 As examples of theories contributing for the understanding of M&A phenomena, the market for corporate control (Jensen & Ruback, 1983), economies of scale (e.g. Jensen, 1993; Lambrecht, 2004), or management selfishness (e.g. Mueller, 1969, 1989), can be mentioned. Nevertheless, none of these studies contributed significantly for the construction of a comprehensive theory.

As referred by Ribeiro & Crowther (2007b), more recent research attempts to explain overall M&A activity using a neoclassical approach (See e.g. Andrade et al., 2001; Andrade & Stafford, 2004; Gort, 1969; Harford, 2005; Mitchell & Mulherin, 1996; Sudarsanam, 2003; Weston et al., 2004), arguing that merger waves result from shocks to an industry’s environment, such as technological innovations or deregulation (Harford, 2005); while other authors believe that M&A waves occur as a result of temporary stock market misvaluation (e.g. Dong et al., 2006; Rhodes-Kropf et al., 2005; Rhodes-Kropf & Viswanathan,

114 Synergy, a vague concept in M&A, comes from the basic notion that the global value of individual parts that come together exceeds the sum of the values of the parts when kept alone, i.e., 1 + 1 > 2. This is not necessarily true, as demonstrated by Roll (1986). In his “hubris hypothesis”, he argued that M&A, namely takeovers in the sense of acquisitions, neither create nor destroy value but simply redistribute wealth from hubris acquirers to target shareholders. The hubris hypothesis relies on the notion that an excessive auto-confidence from the bidder results in a loss for their shareholders due to high acquisition premiums offered. The fate of this type of takeover success, Pirro-like victories that result in later vulnerability of the bidder due to substantial loss of value, has been called the “winner curse”. For a review on M&A success see e.g. Sirower (1997).

127

2004; Shleifer & Vishny, 2003).115 This second approach, commonly labelled behavioural, is built on theoretical and empirical research which has observed a positive statistically significant correlation between aggregate share valuations and merger activity (Beckenstein, 1979; Becketti, 1986; Golbe & White, 1988; Guerard, 1985; Markham, 1955; Melicher et al., 1983; R. L. Nelson, 1959; Steiner, 1975; Weston, 1953).116 Beyond the current mainstream approaches, there are also a number of diverse attempts using other different arguments, but with little impact regarding a contribution for the construction of a comprehensive M&A theory.117

In terms of the type of data utilised, several empirical studies have tested the wave pattern using aggregate industry data (e.g. Barkoulas et al., 2001; Becketti, 1986; Golbe & White, 1993; Melicher et al., 1983; Mueller, 1980; Town, 1992), while others studied this phenomenon at industry (e.g. Andrade et al., 2001; Eis, 1969; Gort, 1969; Harford, 2005; Mitchell & Mulherin, 1996), or at institutional levels (e.g. Auster & Sirower, 2002).

In conclusion, there are many theories but they can provide only partial explanations, based in specific time periods and factors. For instance, a theory that can explain satisfactorily the occurrence of fifth merger wave may not be suitable for the following M&A wave. Despite a comprehensive and widely accepted theory is missing, the many existing partial theories can be used to specifically fit a model that enables to test the research hypotheses, as aimed in this thesis.

115 Market misvaluation can be defined as the discrepancy between the market price and a present measure of the fundamental value (Dong et al., 2006). 116 The wide existent literature is quasi unanimous about it. 117 e.g. Holmstrom & Kaplan (2001) focus on the role of corporate governance in the occurrence of M&A waves.

128

3.4 New age for business combinations and goodwill accounting

As the discussions surrounding business combinations accounting were examined in depth in Chapter 2, in this section the formal FASB’s process and steps that led to the new M&A accounting are summarised, highlighting the main changes to the previous GAAP set on the matter.

Following the ongoing criticism on pooling of interests method in the 1990’s (e.g. M. L. Davis, 1991; Dieter, 1989; Rosenfeld & Rubin, 1985), in August 1996, the FASB added the project on accounting for business combinations to its agenda, with the objective of improving the transparency of accounting and financial reporting of business combinations. In 1997, FASB published a Special Report, Issues Associated with the FASB Project on Business Combinations, (Financial Accounting Standards Board, 1997b), asking for contributions on the scope, direction, and conduct of the business combinations project. In 1998, the G4+1 issued a Position Paper, Recommendations for Achieving Convergence on the Methods of Accounting for Business Combinations (G4+1 (Organization) & International Accounting Standards Committee, 1998). In this document, pooling of interests and purchase methods were both initially considered as well as a third one: the fresh-start method.118 However, G4+1’s final recommendation was that only the purchase method should be used to account for business combinations.

Despite the tightening of pooling usage by the accounting regulators, pooling subsisted over time. Fig. 3.3, constructed based on AICPA data (American Institute of Certified Public Accountants, 1968-2003) shows how the APB Opinion 16’s 12 criteria set for pooling might have worked. The impact on pooling usage was gradual rather than immediate. The fast decrease in the years following the effectiveness of the

118 The concept of “fresh-start” was not new. In fresh-start accounting, the assets and liabilities of the combining companies are recognised in the balance sheet of the combined entity at fair value (not at cost as in pooling). The combined entity is therefore treated as a brand new entity as in a consolidation deal.

129

Opinion was certainly due to the further regulation constraining pooling adoption issued by the accounting standard setters and the SEC. Altogether with other regulation (vid. e.g. Weber, 2004) and with developments in M&A activity, it led to a significantly reduction of the pooling of interests use: from around 50% in the late 1960’s, to around 10% in the 1990’s (American Institute of Certified Public Accountants, 1968-2003).

APB Opinion No. 16 SFAS No. 141 100%

75% Purchase Method Pooling of Interests 50%

25%

0% 7 1 5 9 3 7 1 5 9 * 96 97 97 97 98 98 99 99 99 1 1969 1 1973 1 1977 1 1981 1 1985 1 1989 1 1993 1 1997 1 2001

Fig. 3.3 Pooling versus purchase in the USA, 1967-2001 * From 1st July 2001 data referred only to purchase method, as pooling of interests ceased. Data source: based on a dynamic sample of 600 firms used by AICPA in Accounting Trends & Techniques, several ed. (American Institute of Certified Public Accountants, 1968-2003).

A close observation and interpretation of Fig. 3.3 poses a possibly challenging question for this thesis’ main research topic: the low percentage of pooling adoption in the last years seems to suggest that pooling is rather insignificant in the 1990s and therefore one could wonder whether its abolition could also be insignificant? Despite appealing, this question cannot be, however, sustained by sizeable data on pooling versus purchase usage, and therefore cannot be satisfactorily answered. A brief AICPA’s sample description follows in order to understand why.

130

The 600 companies included in AICPA’s survey are all registered with the SEC, being in recent years selected from Fortune 1000 components list. They are therefore mostly large and listed companies.119 Indeed, when measured by revenue, in 2001, only 18 companies had less than $100 million in revenue, while 491 companies had at least $1 billion in revenue.120 Large listed companies represent much less than 1% in the USA and consequently AICPA’s data cannot be regarded by any means as statistically representative of the population.121 Therefore, data shown in Fig. 3.3 needs to be observed carefully, and may perhaps be used mostly as a trend indicator, and not necessarily meaning that pooling usage was merely residual.

Furthermore, other data sources suggest of a more important role of pooling of interests among companies involved in M&A. In a survey with almost 200 responses, 65% of the respondents admitted they have been through a merger in the previous 10 years (Association for Financial Professionals, 2000). Among those involved in past mergers, 33% of respondents have used either pooling of interests exclusively, or both pooling and purchase methods (Association for Financial Professionals, 2000). Additionally, in Fig. 3.4, also in this section, it is also possible to observe another different data source, which reveals that, in the late 1990s, the total dollar value of M&A deals using pooling of interests was

119 Every AICPA’s sampled company is traded on major USA stock exchanges, or on “over-the- -counter” exchanges (American Institute of Certified Public Accountants, 1968-2003). 120 In 2001, the number of companies in AICPA’s sample with less than $100 million in revenue was 25. In 1998 and 1999, only 23 companies were under this threshold (American Institute of Certified Public Accountants, 1968-2003). 121 There is no official definition of large company in the USA, but taking into consideration the thresholds referred regarding AICPA’s sample selection, and according to the Statistics about Business Size from the U.S. Census Bureau (2008), from a total of 5,697,759 employing companies in 2002, 0.26%, or 15,103 companies, had more than $100 million in revenue, and only 1,947 companies had more than $1 billion in revenue, totalling 0.034%. These figures would be reduced whether nonemployer companies would be included - in this case the total number would include 23,343,821 firms. Additionally, and also according to the U.S. Census Bureau (2008), during the period 1988-2004, there were only 0.3% employing companies with 500 or more employees.

131 higher than in purchase method. In resume, it is clear that the importance of pooling has been reduced since the 1970s, but it is not clear whether it was absolutely insignificant in the 1990s. Despite the existence of some sparse data, it is not possible to draw any authoritative conclusions as the main databases in accounting and finance do not provide information about the accounting method used in M&A deals.

Finally, this thesis studies the impact of changes in business combinations accounting, which includes not only the impact of pooling elimination, but also the impact from the changes in the accounting treatment of purchased goodwill. Therefore, even if pooling of interests would be completely insignificant in the 1990s and in the early 2000’s, this thesis’ research questions would remain meaningful. The remainder of this chapter will help to understand why.

Indeed, although data shown before in Fig. 3.3 suggested that pooling was somewhat irrelevant when FASB proposed its prohibition, only around 10% of the number of deals in the 1990s, it may, nevertheless, provide a biased view.122 According to King & Kelly (2000), measured by dollar value, 55% of business combinations were accounted for pooling of interests. This suggests that some managers involved in megamergers could prefer pooling of interests. However, there is no data or enough evidence that may strongly support the suggestion that the preference for pooling was greater in megamergers than in more modest M&A deals.123

Examples of megamergers using pooling included Daimler-Benz and Chrysler, SBC Communications and Ameritech, Exxon and Mobil, and Citicorp and Travelers Group (Duncan & Carleton, 1999). It may also be of interest to recall that, for example, in the US Senate Banking Committee hearing on pooling accounting, Jim Barksdale, former Chief Executive Officer (CEO) of Netscape, testified that the AOL / Netscape deal would not have taken place if annual goodwill amortisation had been recorded (Prepared Testimony of Mr. James

122 Vid. also the comments made earlier about existing data on poolings vs. purchase use. 123 Ibidem.

132

Barksdale Partner The Barksdale Group, 2000), being large sectors of the business community, particularly from technological areas, actively lobbying against pooling of interests elimination.

Despite the limitations referred to earlier concerning the examination of the pooling versus purchase use issue, Fig. 3.4 is shown below in order to provide a different view about the perceived importance of pooling before its elimination, at least when measured by dollar value, and not only by number of deals.124

100%

75%

50%

25%

0% 1998 1999 2000 Pooling Purchase

Fig. 3.4 Percentage of deals using pooling of interests versus purchase method, in dollar value, 1998-2000 Data source: First Call / Thomson Financial, 2006.

It is also relevant to be aware that there is not comprehensive data available about pooling usage by industries. Nevertheless, anecdotal evidence supports the suggestion that pooling of interests was largely favoured in some industries, such as IT, industrials, and financials, and therefore one can consider the likelihood of a considerable level of pooling usage, at least in some sectors.125

124 Data shown in Fig. 3.5 is merely a sample prepared by First Call / Thomson Financial, as SDC platinum database does not include information about the accounting method used in M&A deals. 125 e.g. Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000; Prepared Testimony of Mr. James Barksdale Partner The Barksdale

133

This is relevant for this thesis, as it reinforces the importance of pooling of interests role, increasing the interest of examining the possible impacts of the changes in business combinations accounting.

In resume, even if pooling of interests usage was reduced, there was something about it that made this method so valuable to the point where companies were ready to defend it as fiercely as shown before in Chapter 2. Whether it is merely a question of megamergers deals with massive goodwill figures it is also unclear, as the inconvenience of the recognition of purchased goodwill stands for every company forced to adopt purchase method.

The FASB kept its determination regarding pooling elimination. Following a long period of discussions and due process, including an invitation for comments (based on the G4+1 Position Paper), public hearings, an exposure draft and a later revised version, and numerous comment letters received, finally the FASB issued two standards in June 2001: SFAS No. 141, Business Combinations, and SFAS No. 142, Goodwill and Other Intangible Assets, superseding APB Opinions No. 16 and 17, respectively (Financial Accounting Standards Board, 2001a, 2001e).126

The provisions of SFAS 141 are effective for all business combinations initiated after 30 June 2001, and for all business combinations accounted for using the purchase method for which the date of acquisition is 1 July 2001, or later. It does not apply to: combinations of two or more not-for-profit organisations; acquisitions of a for-profit business entity by a not-for-profit organisation; and

Group, 2000; Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000; Testimony of Kimberly J. Pinter, Director, Corporate Finance and Tax National Association of Manufacturers On behalf of the National Association of Manufacturers before the Committee on Banking, Housing, and Urban Affairs United States Senate On Business Combinations and Intangible Assets, 2000. 126 The discussions and due process that preceded the adoption of these standards is to be treated in the next section of this chapter.

134 combinations of two or more mutual enterprises. Some significant differences between SFAS 141 and APB 16 are as follows:127

i) SFAS 141 eliminates the pooling of interests method of accounting for business combinations, therefore all business combinations are required to be accounted for by a single method - the purchase method;

ii) SFAS 141 requires intangible assets to be recorded apart from goodwill if they meet one of the criteria established in the pronouncement (either the contractual-legal criterion or the separability criterion), in contrast to APB 16, which required separate recognition of any intangible asset that could be identified and named;

iii) SFAS 141 significantly increases the disclosures about business combinations (e.g. disclosure about the primary reason for a business combination and the allocation of the purchase price to the assets acquired and liabilities assumed. Furthermore, when the amounts of goodwill or intangible assets acquired are significant in relation to the purchase price, it requires disclosure of other information about those assets, including the reporting of the amount of goodwill by segment if applicable SFAS No. 131, Disclosures about Segments of an Enterprise and Related Information, and the amount of the purchase price assigned to each major intangible asset class).

Like APB Opinion 17 and APB Opinion 16, SFAS 142 is intimately linked to SFAS 141 and therefore they share the same rationale. According to SFAS 142 Summary, the standard is to be applied to all goodwill and other intangible assets recognised in the entities’ financial statements (Financial Accounting Standards Board, 2001e). Impairment losses for goodwill and indefinite-lived intangible are due to arise and to be reported. Companies are required to apply SFAS 142 in their fiscal year beginning after December 15, 2001 (early adoption is permitted for entities with fiscal years beginning after March 15, 2001,

127 Adapted from SFAS 141 Summary (Financial Accounting Standards Board, 2001a).

135 provided that their first interim financial statements have not been issued). The major differences between SFAS 142 and APB Opinion 17 are outlined below:128

i) Under APB Opinion 17, the portion of the premium related to expected synergies (goodwill) was not accounted for appropriately. SFAS 142 adopts an aggregate view of goodwill and bases the accounting for goodwill on the reporting units of the combined entity into which an acquired entity is integrated;

ii) APB Opinion 17 assumed that goodwill and intangible assets were wasting assets and therefore considered appropriate to amortise them over their useful lives limited to an arbitrary ceiling of 40 years. By opposition, with SFAS 142, goodwill and indefinite useful lived intangible assets are no longer to be amortised but to be tested at least annually for impairment (or more frequently if necessary). Intangible assets with finite lives will continue to be amortised over their useful lives, but SFAS 142 does not impose any limit;

iii) Lack of guidance in the previous standards related to the determination and measurement of goodwill impairments resulted in diversity in practice. SFAS 142 provides a model and methodology to test and measure goodwill impairment: the two- -step model.129 It also provides specific guidance on testing intangible assets not to be amortised for impairment;

iv) APB Opinion 17 required intangible assets to be amortised using the straight-line method (unless demonstration that other method could be more appropriate). SFAS 142 changes this presumption: intangible assets are to be amortised using a method according the economic benefits status of the intangible asset (straight-line method to be applied only if not possible to determine those benefits);

v) Finally, SFAS 142 requires additional disclosure of information about goodwill and other intangible assets in the years subsequent to completion of the M&A (changes in the carrying amount of

128 Adapted from SFAS 142 Summary (Financial Accounting Standards Board, 2001e). 129 Step 1: test for potential impairment; Step 2: measurement of the impairment (if necessary). Vid. Chapter 6 for more details.

136

goodwill from period to period; the carrying amount of intangible assets by major intangible asset class for those assets subject to amortisation and for those not subject to amortisation; and the estimated intangible asset amortisation expense for the following five years).

To summarise, according to the new FASB’s GAAP, the purchase method is the single accounting method accepted for business combinations, and purchased goodwill is not to be amortised, but to be subject to impairment tests. These statements have been the focus of much controversy and are arguably some of the most significant statements ever issued by the FASB as they resulted in a deep change in the way companies accounted for business combinations, therefore deserving further examination, and also deserving to be subject to testing for possible impacts on M&A activity.

3.5 Neutrality versus economic effects in M&A accounting

As discussed earlier in this thesis, financial reporting is important for the business community and for stakeholders, but in order for it to be useful, it needs to be guided by specific objectives and should also have some critical qualitative characteristics, like the ones mentioned at FASB conceptual framework. Indeed the FASB, in SFAC 1 (paragraph 34, Financial Accounting Standards Board, 1978), states that:

“Financial reporting should provide information that is useful to present and potential investors and creditors and other users in making rational investment, credit, and similar decisions.”.

The main purpose of accounting information is to provide useful information to decision makers. Through SFAC 2, Qualitative Characteristics of Accounting Information, published in 1980b, FASB recognises the usefulness for decision making as the most important characteristic of information. Accordingly, usefulness is placed at the top of the hierarchy of accounting qualities, being considered at the “user-specific quality” level. Below, at the “primary decision-

137

-specific qualities” level, are placed two other critical qualities: relevance and reliability. Directly concerned with reliability is the characteristic of neutrality. According to SFAC 2 (paragraph 98, Financial Accounting Standards Board, 1980b):

“Neutrality means that either in formulating or implementing standards, the primary concern should be the relevance and reliability of the information that results, not the effect that the new rule may have on a particular interest. A neutral choice between accounting alternatives is free from bias towards a predetermined result. The objectives of financial reporting serve many different information users who have diverse interests, and no one predetermined result is likely to suit all interests.”.

Accounting neutrality is not consensually accepted and some reject it because it seems utopian (vid. e.g. Hendriksen & van Breda, 1992). In fact, the accounting has purposes and the accounting information cannot avoid affecting behaviour (paragraph 102, SFAC 2):

“Information that reports on human activity itself influences that activity, so that an accountant is reporting not on some static phenomenon but on a dynamic situation that changes because of what is reported about it.”

Nevertheless, FASB’s mission statement points out that (SFAC 2, paragraph 100):

“To be neutral, accounting information must report economic activity as faithfully as possible, without coloring the image it communicates for the purpose of influencing behavior in some particular direction”.130 131

130 Vid. also FASB Rules of Procedure, p. 3. 131 Original italic.

138

Therefore, what matters is to have information as neutral as possible, even knowing that utterly neutral information may only be possible from a pure theoretical point of view.

If the accounting information could be effectively neutral then the pronouncements would not be supposed to produce economic effects. FASB made a very interesting assertion at SFAC 2 (paragraph 106):

“while rejecting the view that financial accounting standards should be slanted for political reasons or to favor one economic interest or another, the Board recognizes that a standard-setting authority must be alert to the economic impact of the standards that it promulgates.”.

This means that the board recognises that it must be immune, as much as possible, to governmental pressures and entrepreneurial lobbies. Indeed, FASB stresses the importance of its own neutrality for the credibility of any standards it may produce (SFAC 2, paragraph 104):

“But more importantly, it is not desirable for the Board to tack with every change in the political wind, for politically motivated standards would quickly lose their credibility, and even standards that were defensible if judged against the criteria discussed in this Statement would come under suspicion because they would be tainted with guilt by association.”.

As political and corporate influences are to taint, at the very least, the credibility of financial standards, the FASB’s faces, nevertheless, an everlasting need to fight for its independence – threatened by the need of balancing its positions, often being forced to compromise due to corporate and political pressures.

Indeed, it seems generally accepted, particularly from the FASB members viewpoint, that the board processes have to continue to be objective and independent (see e.g. Beresford, 1989, 2001; V. Brown, 1990; P. B. W. Miller,

139

1990; Zeff, 2002). However, although the FASB is represented as having no moral or legal authority to establish public policy and makes no claims to such authority, it needs to be very aware of the social and economic impact caused by its accounting rules (see e.g. Burchell et al., 1980; Hines, 1988; Zeff, 1978).132

One could argue that accounting policy should be neutral and should not produce any economic effects. However, accounting standards do have social and economic consequences (Burchell et al., 1980; Hines, 1988; Zeff, 1978), as FASB explicitly admits it in SFAC 2. The minimisation of these consequences is therefore a main concern.

It is never easy to predict and isolate those potential effects from standard- -setting policy from other economic events, especially in the case of new standards, where the estimation is less reliable. In case an unexpected and undesirable effect on business behaviour is generated, it may then be necessary to review the accounting standards (vid. e.g. Hendriksen & van Breda, 1992).

The business combinations topic is very sensitive concerning the neutrality of accounting information. The M&A market is very dynamic and important in the USA, the deals can involve massive dollar values, and the activity is considered critical for economic restructuring, development, and growth (vid. e.g. Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000; Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, 2000). Without surprise, FASB recognised that the standards it would produce should not seek to encourage or discourage business combinations (paragraph B75, SFAS 141):

“Instead, those standards should produce information about those combinations that is fair and evenhanded to those having opposing economic interests.”.

132 Vid. also other literature about social and economic consequences of accounting policy (e.g. American Accounting Association, 1978; Benston, 1978; Bromwich, 1985; Horngren, 1973; Kelly- Newton, 1980; Moonitz, 1974; Rappaport, 1977; van Riper, 1994; Wyatt, 1977; Zeff, 1993).

140

The board also exploited the neutrality attribute to support the pooling elimination. As referred earlier, pooling of interests’ supporters argued that the M&A activity could be harmed if the FASB would not change its proposals. According to the FASB, such defence could mean that pooling encouraged certain types of business combinations, therefore suggesting that the existing business combinations accounting was not neutral (paragraph B75, SFAS 141):

“In the final analysis, the Board concluded that the accounting standards for business combinations should not seek to encourage or discourage business combinations.(…) The Board also concluded that those who argue for the pooling method on the basis that they believe that it fosters more combinations are not seeking to have neutral, evenhanded information disseminated.”.

It can be said that FASB was sharp enough to use the main argument from pooling defenders as a justification for pooling elimination. Indeed, it is suggested that the previous business combinations accounting standards continued to produce economic effects, meaning that the pronouncements issued by the standard setters over time were not enough to significantly mitigate the distortions caused by pooling adoption. Moreover, it can also be argued that the successive restrictions imposed over time to limit the possibility of pooling adoption increased the level of distortion that companies were willing to accept in order to qualify M&A deals as unity of interests. As to be discussed later in this thesis, anecdotal evidence supports the idea that such distortion could involve the artificial reshaping of M&A deals, often involving large amounts of reshaping costs (vid. e.g. the case of AT&T’s acquisition of NCR).

When looking for the best accounting treatment for goodwill, where several different acceptable methods have been initially considered, the FASB (paragraph B98, SFAS 142, Financial Accounting Standards Board, 2001e):

“did not consider it possible to develop a method of accounting for acquired goodwill that all would agree established a level playing field in all circumstances. Accordingly, the Board focused on which

141

method better reflects the economic impact of goodwill on an entity.”.

Moreover, (paragraph B99, SFAS 142):

“the Board reaffirmed its decision that non-amortization of goodwill combined with an adequate impairment test will provide financial information that more faithfully reflects the economic impact of acquired goodwill on the value of an entity than does amortization of goodwill.”.

The FASB’s views on neutrality, and on economic consequences from accounting policy changes, would be reflected in the process of revision of the business combinations accounting which started in the mid-1990’s. However, FASB would face strong opposition, being the board’s independence threatened, as examined in chapter 2. Like in the late 1970’s, many finance professionals and managers argued that if pooling was to be discontinued, the M&A activity would suffer, resulting in a negative impact on the USA economy (inter alia King, 2000; King & Kelly, 2000; Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000; E. E. Smith, 1999). It was also referred that the lobbying from the business community was so intense to the point that politics became directly involved in the FASB’s business combinations project, endorsing hearings in the US Congress and with involvement from the US Senate (see e.g. Beresford, 2001; Zeff, 2002). Nevertheless, despite the political and economic pressures to keep both methods, much of the literature in accounting kept defending the abolition of pooling of interests (see e.g. American Accounting Association. Financial Accounting Standards Committee of Accounting Procedure, 1999).

As much inconvenient a change in GAAP can be, the fact is that evolution is inevitable in accounting (vid. e.g. Zeff, 2003a, 2003b). If change is inevitable, then economic consequences from changes in accounting policy may be as well (vid. e.g. Hendriksen & van Breda, 1992). As addressed in this research, whether the economic consequences from changes in accounting policy may have resulted inconvenient is the crux question. In the next section, an early analysis

142 of the possible impacts on M&A activity following FASB’s changes on business combinations accounting is provided.

3.6 Possible impacts on M&A activity and M&A accounting

The discussion about the need for a change in business combinations accounting seems to have been fruitful. Following the effectiveness of the new standards, the dissidents of pooling could finally end their criticism. More importantly, the worst fears of the defenders of pooling appear not to be confirmed. Looking at Fig. 3.5 below, it is possible to observe that the M&A market stabilised after July 2001, following the early 2000’s exchange markets crash and the end of the fifth M&A wave. There is no clear indication of a last minute rush for pooling qualification, and the M&A market remained flat in 2002, in line with the subsisting business environment.

Jul 2001 = 100

500 150 400

300 100

200 50 100

0 0 00-1 00-5 00-9 01-1 01-5 01-9 02-1 02-5 02-9 03-1 03-5 03-9 04-1 04-5 04-9 05-1 05-5 $ Value (left scale) No. of Deals (right scale)

Fig. 3.5 Monthly M&A completed deals in number of deals and dollar values in the USA, from 01-2000 to 05-2005 Data source: SDC Platinum (Thomson Financial, 2006).

143

In terms of the number of companies accounting for business combinations, the figures appear to have continued stable, despite the pooling elimination and the changes in purchase method. For example, according to Accounting Trends & Techniques (American Institute of Certified Public Accountants, 1968-2003), in 2002, 314 of the 600 surveyed companies used the new purchase method to account for business combinations, versus 309 purchases and 16 pooling of interests in 2001.

Additionally, apparently there were not significant indications of relevant economic effects produced by the new standards. While the neutrality of the new standards seems not to be an issue, not everything remained necessarily unchanged. Otherwise, the FASB changes would not be more than merely cosmetic and would not produce any effects. That would be abnormal. As Hendriksen & van Breda (1992: 242-243) note:

“All decisions regarding accounting policy should have economic consequences. If there were no consequences, there would be no reason for policy decision. The desired consequences include an improvement in the information available to investors and other users with the result of permitting sounder economic decisions or a reduction in the information-gathering costs for users. Through the securities markets, better decisions should result in an allocation of resources closer to the optimum and an opportunity for an improvement in portfolio selections. If decisions are not altered and if information costs to users are not reduced, this is evidence that the policy decision was not desirable”.

Therefore, the appropriateness of the changes in business combinations and goodwill accounting could be simply judged by the absence of undesirable effects. Furthermore, one could argue that the changes should be useful, resulting in positive effects for the users, particularly investors. In order to examine whether FASB’s changes produced any significant effects on M&A activity, a set of hypothesis are formulated in chapter 4, being subject to testing in chapter 7, using different sets of samples as shown in chapter 6.

144

Besides any possible effects of the new GAAP on the M&A activity, which are to be tested later in this thesis, the new standards carried some major changes about the way companies looked to business combinations accounting. The most important one is the elimination of the goodwill amortisation charge which increases the value of reported earnings and EPS, a fundamental ratio for managers and financial analysts. Naturally this will be true as long as the company does not recognise any loss by impairment. Therefore, the GAAP change carried a possible relevant impact on financial reporting. As an example, the table below compares the first time application of SFAS 142 versus a pro forma reporting under APB Opinion 17 for the same period.

Table 3.1 Pro forma impact on EPS in 2001 of selected S&P 500 companies Goodwill APB SFAS Percent Amortisation Op. 17 142 Change Company Industry Classification Expense EPS EPS EPS General Electric Co. Industrial Conglomerates $1,100 $3.22 $3.42 6.2% Pfizer, Inc. Pharmaceuticals 20 0.82 0.82 0.4% Intel Corp. Semiconductors 307 2.20 2.26 2.5% Merck & Co. Pharmaceuticals 365 2.51 2.60 3.7% Johnson & Johnson Pharmaceuticals 93 3.00 3.04 1.3% Philip Morris Tobacco 582 3.21 3.36 4.6% Berkshire Hathaway Property/Casualty Insurance 477 1025.0 1213.4 18.4% Tyco International Industrial Conglomerates 344 2.68 2.80 4.6% Wells Fargo & Co. Major Banks 530 2.36 2.55 7.9% American Home Products Pharmaceuticals 284 -0.69 -0.59 13.9% Data sources: SEC fillings, EDGAR Online

Following the business combinations accounting rules change, companies which had higher goodwill amortisation values before 2001 are likely to be more benefited, experiencing larger pro forma percent changes in EPS. Such impacts are to be studied in depth in chapter 5.

It seems arguable that the new FASB rules added transparency to the business combinations accounting, as accounting choice has been eliminated. This change

145 is also likely to have reduced the cost of analysing financial reporting. In the past, analysts placed lower valuations on companies that amortised goodwill using the purchase method than on companies which did not amortise goodwill, either because they were using pooling, or because they preferred to write off the bid premium in the first year as in-process research and development expenses (Hopkins et al., 2000). As a result, the pooling ban eliminates the bias that analysts may have priced into stocks of pooling companies.

Perhaps the major drawback of the business combinations accounting changes is the use of goodwill impairment tests. Regarding purchased goodwill, and other intangible assets, companies are expected to recognise losses by impairment, whenever the deals have not met expectations. However, despite the available guidance about the developing of impairment tests, the decision about undergoing impairments is possibly subject to managerial discretion. Therefore, the new rules introduced an element of uncertainty and additional risk into future earnings estimation due to potential impairment surprises. Nevertheless, stock prices should not decline significantly for companies with one-time impairment write-offs unless they become habitual (Hopkins et al., 2000). Together with the analysis of the impacts on EPS, the analysis of possible effects from impairments is also to be examined in chapter 5.

3.7 Conclusions

One of the main arguments used in favour of pooling of interests was the possible negative impact from its elimination on M&A activity, and consequently on the economic activity. As FASB recognises, the accounting policy produce effects, and therefore any changes in business combinations accounting needed to be carefully discussed, as M&A is a very important activity for the economy of the USA. The FASB continued arguing that the pooling elimination would end accounting choice in business combinations accounting, and therefore the new standards would bring more transparency and fairness of financial reporting to the companies involved in M&A deals. In

146 order to achieve this desiderate, the FASB had to compromise and to find a conciliatory solution: pooling was definitely to be eliminated, but purchased goodwill accounting would be changed. Indeed, replacement of purchased goodwill amortisation by impairment testing seemed to have mitigated the outstanding criticism from the business community (Ribeiro & Crowther, 2008b).133

Whether FASB’s changes did not produce any economic effects on M&A activity is not clear. Apparently, there were no significant economic effects on M&A activity and on the economy in general. Indeed, the M&A activity did not suffer any significant setback and the economy recovered from the weak environment of the early 2000’s. Nevertheless, as shown in Fig. 3.5, the M&A activity remained flat for a long period, and therefore one can wonder whether it may be related to FASB’s changes. Moreover, the examination of the changes in business combinations accounting and its possible impacts on M&A activity and on financial reporting are yet to be comprehensively examined by the literature, being therefore these issues to be the subject to an in depth exploration in this thesis.

133 To best of the knowledge of this author’s thesis, systematic criticism about pooling elimination could not be found expressed in more recent literature.

147

Chapter 4 Hypotheses

4.1 Introduction

As examined earlier, by means of literature review and analysis of accounting standards, changes in accounting policy may produce economic and social effects. The forces and actors behind such changes were also subject to examination. In fact, there is a perceived dichotomy between standard-setting accounting policy, and corporate and political lobbying. The case of the changes in business combinations accounting illustrates such dichotomy well.

Another reason for making business combinations accounting such an interesting phenomenon to study, was the controversial duality that prevailed until recently in the USA. As discussed earlier, the accounting choice in business combinations has always been subject to criticism (e.g. Briloff, 1967, 1968; M. L. Davis, 1991; Eigen, 1965; L. C. Phillips, 1965; Rosenfeld & Rubin, 1985; Wyatt, 1967), and its effects in financial reporting and in managers and investors decisions are well documented in literature (e.g. Aboody et al., 2000; Anderson & Louderback III, 1975; Ayers et al., 2000, 2002; R. M. Copeland & Wojdak, 1969; M. L. Davis, 1990; Erickson & Wang, 1999; Gagnon, 1967; Louis, 2004; Nathan, 1988; Weber, 2004).

Like business combinations accounting, the study of the occurrence of M&A waves, a stream of research from economics and finance, is also of interest for this thesis, as it is a major phenomena in M&A activity. As examined earlier in

148 this thesis, although broadly researched, it was not, however, possible to develop a comprehensive theory to explain its occurrence. Nevertheless, there is plenty of evidence concerning the M&A activity, providing diverse explanations about its occurrence and pattern, which contributes in terms of M&A understanding, being therefore valuable for the developing of the present research.

Having discussed earlier the economic consequences from accounting policy changes, it is therefore of interest to explore whether the accounting for business combinations changes have affected the M&A activity, measured by the number of deals completed.134 This is the first main research hypothesis to be tested.

As many have argued that the proposed changes in M&A accounting policy would constrain M&A activity, these are also examined in this chapter, by means of literature review on M&A cancelations and withdrawals, whether such suggestion should also be taken into consideration. As it was found feasible to test such suggestion in a form of an assumption, a second main research hypothesis was formulated, as shown in this chapter.

The presentation of the research hypotheses is preceded by an in depth description of the M&A activity during the periods of study: short and medium term, in the period 2000-2002; and long term, during the period 1994-2008. Such description is followed by the presentation of the main research methodology used in M&A empirical studies. Such exhibition is used to justify the methodology that has been chosen to test the research hypotheses.

Finally, after the presentation of the research hypotheses and their underlying rationales, the reasons that justify the development of the hypotheses in the current form are shown in the conclusions.

134 It could also be of interest to measure by the value of M&A deals completed. However, as to be point out later in this thesis, in most cases M&A deals values are not disclosed and, when so, valuations are often not consistent among the several M&A data providers, particularly whenever payment includes other means than cash (e.g. shares, options, warrants).

149

4.2 M&A activity during the 2000-2002 period

In order to assess any potential effects on the M&A activity as a consequence of the changes on the accounting regulation, it may not be advisable to use a very short period of a few days only, like many studies on M&A returns, as the effects may last for larger periods: weeks, months, or even years. On the other hand, nor should one choose to have a very long period, like studies on M&A waves which necessarily make very long term analyses, as such effects may be totally diluted in very large periods. However, impacts may have persisted hitherto and may therefore be visible in the long run. Accordingly, a long term analysis is also of interest and has therefore been done in this research, time obviously being limited by the largest existing dataset available when this research has been developed. As a result, this study has a long-term period of analysis - fifteen years, from 1994 to 2008 – which should be long enough to capture any long lasting effects; and a short-middle range period of analysis - three years, from 2000 to 2002 - where any possible short and middle term impacts are more likely to be revealed.135

During the triennial period started in 2000, the overall M&A activity in the USA, both in number and value of deals, was of a general downward trend, as shown previously in Fig. 3.5. A similar scenario was set for stock exchange markets. On 14 January 2000, the Dow Jones index started a 33-month slide, and eight weeks later NASDAQ would follow in its steps. In 2000, it was not only the stock markets that started a bearish period, since the positive economic cycle and the M&A activity were fading as well.136 The M&A activity slide then

135 The middle range period of analysis was defined to be 3 years, as this corresponds to the middle of the interval of what can be considered a short term period - less than one year - and a long term period - more than 5 years. Since such selection can be regarded as arbitrary, a sensitivity analysis was also developed in order to cover shorter period intervals. 136 According to the National Bureau of Economic Research (NBER), the economic cycle faced a contraction from March to November of 2001 (National Bureau of Economic Research, 2007). It is important to note that NBER does not necessarily recognise a period of two consecutive

150 reached a bottom by the time of the September 2001 terrorist attacks. Following the disruption caused by the 9/11 events, the activity started a stagnation period that would prevail until the end of 2002.

ED FAS 141 & 142 Effective 60

0 3.1.00 3.2.00 3.3.00 3.4.00 3.5.00 3.6.00 3.7.00 3.8.00 3.9.00 3.1.01 3.2.01 3.3.01 3.4.01 3.5.01 3.6.01 3.7.01 3.8.01 3.9.01 3.1.02 3.2.02 3.3.02 3.4.02 3.5.02 3.6.02 3.7.02 3.8.02 3.9.02 3.10.00 3.11.00 3.12.00 3.10.01 3.11.01 3.12.01 3.10.02 3.11.02 3.12.02 Fig. 4.1 Number of announced M&A deals, in the period 2000-2002, during weekdays Order two polynomial trend line added. Data source: SDC Platinum (Thomson Financial, 2006).

Looking at the graph exhibited in Fig. 4.1, one can observe that, apart the initial downtrend, the M&A activity appears remarkably stationary afterwards. The M&A activity also seems to reflect the existing bear stock market, which started in 14 January 2000 and ended in 19 September 2001. The figure signals two of the most important moments in this research: the first being the publication of the revised exposure draft, which confirmed the pooling ban and replaced the initial proposal of purchased goodwill amortisation over a maximum of twenty years, by impairment tests, dated 14 February 2001; and the second being the effectiveness of the final pronouncements, 1 July 2001. The downward trend continued after these events, as one can consider the period of the 9/11 events, and the following end of the bear market, as the reversion quarters of decline in real GDP as a recession, as it rather considers a contraction as “a significant decline in economic activity spread across the economy, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales.”(National Bureau of Economic Research, 2007).

151 point.137 Therefore, one can pose the question as to whether the shift was linked exclusively to endogenous factors, mainly concerning the M&A environment itself, or if it can also be explained by exogenous causes, which includes any possible effects that may have arose from the changes in the accounting rules, changes in stock markets conditions, etc.

0 2/Jul 3/Jul 5/Jul 6/Jul 9/Jul 10/Jul 11/Jul 12/Jul 13/Jul 04/Jul 18/Jun 19/Jun 20/Jun 21/Jun 22/Jun 25/Jun 26/Jun 27/Jun 28/Jun 29/Jun * 7/Jul * 8/Jul * 14/Jul * 15/Jul Ψ * 16/Jun * 17/Jun * 23/Jun * 24/Jun * † 01/Jul * † 30/Jun Fig. 4.2 Number of announced M&A deals in the thirty days surrounding the event date * Weekends Ψ Independence Day (Federal Holiday and non-trading day) † One day event (-1/+1 day standards’ effective date) Data source: SDC Platinum (Thomson Financial, 2006).

An even more in-depth analysis of M&A activity by the time the accounting standards were enforced is made possible through examination of Fig. 4.2. It provides a further zoom into the M&A activity during the two weeks that preceded the effective date, and during the first two weeks thereafter. At first glance, one cannot perceive any unusual circumstance. The first two days after the effective date were the busiest during the period, but that could be explained by a number of other different reasons such as: i) M&A activity tends

137 Reversion point in the sense of a rebound from the 9/11 week low, which generated an immediate positive response, followed by a feeble recuperation, albeit almost imperceptible and that would not last for a long time.

152 to peak whenever a month changes, which is the case as the effective date is 1st July; ii) that tendency to peak may be reinforced in specific months, or whenever a new quarter begins, which is again the case as the date also stands for the start of the third quarter and, cumulatively, second semester of 2001; and iii) the first week of July has a holiday on a Wednesday, which is a non- -trading day, and consequently results in a natural concentration of deals in the beginning of the week.138 Furthermore, the Friday before the key date also seems to be abnormally busy, as Friday is usually the quietest weekday. Actually, the whole week before the critical weekend was good in terms of M&A activity. Once more, the explanation can be related with diverse endogenous factors, such as the ‘end-and-beginning-of-the-month phenomenon’. Further issues related with the peculiar M&A pattern will be later subject to a comprehensive analysis.

Consequently, one could argue that it would not be possible to identify any obvious short-term effects, either positive or negative, resulting from the event in analysis. The activity registered in the first two days that followed the standards effectiveness date could perhaps be explained by the pattern of the M&A activity itself, while the following decrease in the remaining days of the week, could be possibly justified by the mid-week holiday, which resulted in a concentration of deals in the first two trading days of the week. However, there is a fact that casts a shadow over this rationale: usually weekends are extremely flat in terms of M&A activity, regardless any seasonal pattern effects, as to be discussed in the following section (vid. e.g. Branch et al., 2001). weekends analysis: It has been highlighted that weekends are flat in M&A announcements. Indeed, underlying theory related to the timing of M&A announcements supports that M&A agreements are more likely to be completed over a weekend and

138 Conclusions drawn by the author primarily from data observation, and secondly from literature.

153 announced only on the following Monday or during other weekdays (Branch et al., 2001).139

In fact, evidence corroborated for M&A activity during weekends can be accounted as non-existent to residual (Thomson Financial, 2006). For example, according to Thomson Financial (2006) data, at the weekend before the event, not one single M&A deal has been announced. However, the weekend of the effectiveness’ date of the new M&A accounting standards, 1st July 2001, reveals an abnormal activity, as shown below in Fig. 4.3. During this specific weekend, the M&A activity was as high as in an ordinary weekday (vid. again Fig. 4.2), without having anything available in the existing literature about the M&A pattern of activity that could be used to justify such a high level.

1Q 2001 2Q 2001 3Q 2001 4Q 2001 40

0 67676767676767676767676767676767676767676767676767676767676767676767676767676767676767676767676767676767

Fig. 4.3 Number of M&A announcements during weekends in 2001 Number six for Saturdays and number seven for Sundays. Data source: SDC Platinum (Thomson Financial, 2006).

139 This is related to the fact that M&A announcements typically require boards of directors’ approvals (vid. e.g. Branch et al., 2001; Gaughan, 2002; Weston et al., 1998). In order to allow both firms’ board members and supporting staff involved in the M&A deals to consider and discuss the deal terms (e.g. legal and financial advisors, accountants and tax advisors, auditors, etc), this type of operations are usually prepared and completed during weekends, a more quiet period when board members are more free from their businesses duties and have a more released agenda (Branch et al., 2001).

154

The abnormal activity in the -1/+1 event day window is made clearly visible at Fig. 4.3. In 2001, a total 176 deals were announced during the 52 weekends, with more than one third being announced during a single weekend, the event one, with 21 deals in 30th June, Saturday, and 39 deals in 1st July, Sunday (Thomson Financial, 2006). Although these figures benefited from the coincidence of several positive factors, such as the ‘end-and-beginning-of-the- -month phenomenon’, it seems obvious that any global justification for such a high level of activity in a single weekend needs to include the effectiveness of the new accounting standards as an explaining factor.140

Indeed, the effects directly related to the specific M&A pattern of activity may only serve as a partial explanation. Perhaps even as a very minor justification. Serving as illustration, one can take into examination the second busiest weekend in 2001, which matches the end of the first quarter and the beginning of the second quarter. This weekend produced 20 deals, equally distributed by Saturday and Sunday, which, on the one hand, contrasts with the mere three deals average for weekends during the sample period, but, on the other, totals only one third of the activity registered during the event weekend. In addition, apart the event weekend, during the 2000-2002 period, the maximum number of announcements in a single weekend day was on 1 July 2000, Saturday, with 21 deals, at a time when the M&A activity was notably stronger.141

It is therefore arguable that abnormal activity occurred around the event weekend, as the M&A pattern of activity, with its own particular effects, can, at best, provide a partial explanation only. Consequently, it seems reasonable to consider that the effectiveness of the new standards had at least an immediate positive impact on the M&A activity, both previously and immediately during the event day. Previously, as a very unusual M&A activity may be explained by a last day rush for pooling, on 30 June 2001; and immediately, on 1 July 2001,

140 Further issues related with the M&A pattern of activity will be discussed later in this thesis. 141 The weekend of the 1st and 2nd July, totalled 21 deals versus the 60 deals of the event weekend.

155 as then a number of deals, that were possibly delayed to take advantage of the new rules of accounting for goodwill, could finally be announced.

Nevertheless, one could still doubt the reasonableness of the suggestion that the accounting rules change could be a reason for an increase in M&A announcements in the event weekend. Indeed, it may seem reasonable to assume that it is the completion date of the deal, and not the announcement, that would matter in terms of accounting treatment. Therefore it would not be reasonable to focus on the announcements around the event day when, in terms of the accounting rule, the date of completion is what would matter. However, this is not true, because it is the announcement date that is critical. Regarding the effective date and transition, paragraph 59 of SFAS 141 states that (Financial Accounting Standards Board, 2001a):

“(…) this Statement shall be effective as follows:

a. The provisions of this Statement shall apply to all business combinations initiated after June 30, 2001. Use of the pooling- -of-interests method for those business combinations is prohibited.

b. The provisions of this Statement also shall apply to all business combinations accounted for by the purchase method for which the date of acquisition is July 1, 2001, or later.”.

Accordingly, if a deal is “initiated” on 1 July 2001, it can no longer be considered for pooling of interests and therefore has to be recorded under the “new” purchase method, the single method available from the event date. It is therefore suggested that it is the announcement date that it is important, making the completion date utterly irrelevant. As the terms and conditions of “initiated” could result dubious, further clarification was needed, for which the “old” AICPA’s APB Opinion 16 revealed to prove suitable enough for FASB (SFAS 141, paragraph 59, Financial Accounting Standards Board, 2001a):

156

“The following definition of initiated from paragraph 46 of Opinion 16 shall be used in determining the effective date of this Statement:

A plan of combination is initiated on the earlier of (1) the date that the major terms of a plan, including the ratio of exchange of stock, are announced publicly or otherwise formally made known to the stockholders of any one of the combining companies or (2) the date that stockholders of a combining company are notified in writing of an exchange offer. Therefore, a plan of combination is often initiated even though consummation is subject to the approval of stockholders and others.”.

It is therefore clear that it is the M&A announcement date that matters, provided that includes some major deal terms, i.e. that it is not simply a rumour. This fact reinforces the analysis made in this section about the possible importance of the event weekend in terms of announcements. In addition to the considerations already made, an analysis of the eight weeks around the event day also suggests that such positive impact may not be limited only to the weekend event, as it is likely to have been spread onto the immediate surrounding weekdays (Thomson Financial, 2006). It is possible then to conclude that some directly related impact is likely to have been made and one could estimate it as resulting, at least, in some dozens of deals in a three days event window (-3,+3). These figures reveal the existence of an impact in the immediate term, albeit longer-running abnormal effects are not possible to distinguish. However, it is admissible to question whether the impact was also made longer-lasting, affecting the M&A activity not only in the immediate run, but in the short, middle, and long terms as well.

4.3 M&A activity during the 1994-2008 period

In a broader time-interval analysis, between 1994 and 2008 the M&A activity exhibits a cyclical pattern of occurrence, composed during this period by two M&A cycles. Shown below is Fig. 4.4, which documents the fifth M&A wave

157 occurred in the last decade of the 20th century, and also captures the following M&A movement occurred in the 2000’s - which more than a wave in the terms described previously in this thesis, may be considered instead as a common M&A cycle, or, at best, perhaps as a quasi-wave.142 The last section of the graph also documents the effects of the beginning of the latest financial crisis, which resulted in a significant decrease in the M&A activity, with a sudden crash in the fourth quarter of 2008 (4Q08) following the Lehman Brothers’ failure and the financial system quasi-meltdown in September 2008 (vid. e.g. Evans, 2008). The effects of the financial crisis were so dramatic in 4Q08 that it resulted in the lowest level of quarterly M&A activity during the whole period of analysis.

3,000

2,500

2,000

1,500

1,000

500 1Q94 3Q94 1Q95 3Q95 1Q96 3Q96 1Q97 3Q97 1Q98 3Q98 1Q99 3Q99 1Q00 3Q00 1Q01 3Q01 1Q02 3Q02 1Q03 3Q03 1Q04 3Q04 1Q05 3Q05 1Q06 3Q06 1Q07 3Q07 1Q08 3Q08

Fig. 4.4 Quarterly number of M&A deals announced and completed during the 1994-2008 period Data source: SDC Platinum (Thomson Financial, 2009).

It is also interesting to observe that the second lowest level of M&A activity during the 1994-2008 period occurred in 3Q01, exactly the first quarter where any effects from the effectiveness of the new M&A accounting rules could have been made visible. During the fifth M&A wave, the M&A activity peaked in 1997-1998 and again in 1Q00. Then the M&A activity decreased continuously

142 To the best of the author’s knowledge, the existing literature does not broadly and unanimously recognise the M&A activity during the 2000’s as a M&A wave.

158 throughout the time when pooling of interests was still an option and purchased goodwill was subject to amortisation charges. Immediately after the pooling elimination and the replacement of amortisation by impairment tests of purchased goodwill, in 3Q01 the M&A activity recorded a sudden drop, and bottomed. It seems therefore reasonable to continue questioning whether the changes in business combinations accounting also contributed to such decrease of activity.

Additionally, the M&A activity remained lethargic after the effectiveness of the new accounting rules. Indeed, despite a gradual recovery in the quarters following the effectiveness date, the pre-3Q01 level of M&A activity was only restored more than two years after, in 4Q03, and only from 2005 onwards the M&A activity remained systematically above pre-3Q01 levels.143 Therefore, one can reinforce the interest of questioning as to whether the new M&A accounting had any significant contribution to such low level of M&A activity, in this case in the scope of a long period examination.

Finally, data shown in Fig. 4.4 suggests that the publication of the business combinations accounting final ED, in 1Q01 may have not resulted in any immediate significant impacts on M&A activity. Nevertheless, possible medium and long run impacts from the ED issuance need to be taken into consideration, as the level of M&A activity pre-1Q01 was only restored four years later, in 1Q05.

4.4 Research methodology

Regression is the most used research methodology in M&A empirical studies. There are many different types of regressions and modified regression-based models, which have been used for testing hypotheses in M&A research. Some examples follow:

143 Except in 4Q01.

159 studies on M&A waves and pattern of activity: Since early times, several authors utilised correlations and regression models to try to determine and to explain the occurrence of M&A (Markham, 1955; R. L. Nelson, 1959, 1966; Weston, 1961). More recently, a large number of studies, which focused on the M&A waves occurrence, have employed time series and regression-based models. The most common approach was to model M&A series by autoregressive processes, such as ARIMA processes, (e.g. Barkoulas et al., 2001; Chowdhury, 1993; Clark et al., 1988; Shughart II & Tollison, 1984), but other approaches, such as the multi regression model (e.g. Steiner, 1975), or the use of regression analysis to fit a set of sine curves to the time series data on M&A activity (Golbe & White, 1993), were used as well. Nonlinear time series models and Markov regime switching regression models have also been employed (e.g. Linn & Zhu, 1997; Town, 1992).

Currently, the authors who research the M&A pattern of activity continue to apply regression-based techniques, regardless of the research stream. This is reflected in the two mainstream approaches: neoclassical (e.g. Andrade et al., 2001; Harford, 2005; Jovanovic & Rousseau, 2002), and behavioural (see e.g. Rhodes-Kropf et al., 2005; Rosen, 2006), as they frequently use all sort of different types of methodologies involving regressions, from simple ordinary least squares (OLS) estimators, up to event studies, probit, or logit regressions, using the maximum likelihood estimation. studies on returns from M&A announcements: Mostly using standard event study methodologies, in this strand of studies regressions are widely employed to measure cumulative abnormal returns (CAR), a procedure where a regression of the post-M&A abnormal performance on the pre-M&A abnormal performance is applied (inter alia Banerjee & Eckard, 1998; DeLong, 2003; Hazelkorn et al., 2004; Jensen & Ruback, 1983; Leeth & Borg, 2000; Mitchell et al., 2004; S. B. Moeller et al., 2005).144 Additionally, other types of regression models, such as multiple regressions, or logit regressions, are also commonly used to control and examine other issues

144 The event study methodology use goes back to Fama et al. (1969).

160 concerned to M&A announcements and returns (see e.g. Dymski, 2002; Hubbard & Palia, 1999; Matsusaka, 1993).

The event study methodology is relevant to this thesis as it inspired the development of models designed to try to capture abnormal effects based on equidistant time intervals of analysis, ex ante et ex post from a triggering point, i.e., the consideration of the effectiveness of the new M&A accounting rules as an event. studies on M&A accounting methods: Another example of research, which also uses regression models, is the one that examines the effects of the existence of different accounting methods for both M&A deals and purchased goodwill (e.g. Aboody et al., 2000; Ayers et al., 2002; M. L. Davis, 1990, 1996; Hong et al., 1978; Lindenberg & Ross, 1999; Moehrle et al., 2001; Norris & Ayres, 2000; Robinson & Shane, 1990; Weber, 2004). In this type of literature, it is possible to observe the use of CAR regressions, multiple regression analysis, or logit regressions.

This strand of research was also inspirational for this thesis, as it uses methodologies developed to test issues related to the use of pooling versus purchase method, and also related to purchased goodwill amortisation versus impairment testing. This type of studies has also contributed to this thesis’ models development by offering substantive evidence, such as regarding the analysis of the effect of passing time on business combination accounting. studies on M&A cancellations: Finally, the studies regarding M&A withdrawals also use regression techniques, such as: event studies (CAR), OLS estimations, or logistic regression analysis (see e.g. Davidson III et al., 2002; Muehlfeld et al., 2007).

Although the current research is based on existing theory, it nevertheless presents some singular characteristics and poses research questions, which find no parallel in the literature that has been reviewed. The present study has the

161 specific purpose to investigate the existence of any impact on M&A activity as a consequence of the new FASB’s business combinations standards, which abolished pooling and replaced purchased goodwill amortisation by impairment testing. This is in contrast to studies on M&A waves, which try to verify the existence of waves; studies on M&A returns, which use the CAR methodology and are focused on the measurement of market returns provided by announcements; or studies on M&A accounting, which typically look for the pooling versus purchase question, or other issues concerning purchased goodwill amortisation versus impairment.

To summarise, existing literature offers critical findings, valuable methodological bases, and models that were used in order to develop the current research. Additionally, it seems justified to use regression analysis to test the research hypotheses, as regression is the most widely used methodology in mainstream literature in M&A research. Accordingly, the development of the models used to test the research hypotheses are based in regression analysis, as shown in chapter 7.

Besides specific models developed to test the research hypotheses, are also analysed other possible collateral effects from the changes in the accounting policy. Evidence supporting such possibility is presented in chapter 5, by means of analysis of questionnaires and annual reports from companies involved in M&A, following the business combinations accounting changes. The examination of such evidence is then to be triangulated with the results from the regression models’ results, being jointly discussed in the final chapters of this thesis. A discussion regarding methodological issues related to questionnaires and analysis of annual reports in M&A research follows. studies on M&A questionnaires and financial reporting surveys: The use of questionnaires is common in M&A research, and topics may vary widely, from assessment of preferred business combinations accounting method (e.g. Association for Financial Professionals, 2000; K. Nelson & Strawser, 1970); to topics related to post-M&A integration and success evaluation (vid. e.g.

162

Coopers & Lybrand, 1993; Financial Accounting Standards Board, 2000; J. Hunt et al., 1987); among others.

In terms of surveys in accounting and finance related to M&A questionnaires, methodologies also involve regression analysis. In this case, surveys serve mostly as a source of evidence to be used in order to construct variables integrating the development of regression models (vid. e.g. Duggal & Cudd, 1998; Moehrle et al., 2001).

Other methodologies than regression are very common in M&A surveys. Apart testing differences in means between study groups, using Student’s t-test and other parametric statistics, the most common methodology is the use of tests of hypothesis of independence among variables using chi-squared statistic (Brenner & Shuey, 1972; Hopkins et al., 2000; K. Nelson & Strawser, 1970), and also using other nonparametric tests, such as Kruskal-Wallis and Mann-Whitney U (vid. Hopkins et al., 2000).

As for surveys of annual reports concerned with disclosures related to business combinations accounting, there is no particular methodology used, being the methodological analysis often limited to the examination of descriptive statistics (vid. e.g. American Institute of Certified Public Accountants, 1968-2003; Duvall et al., 1992; Wyatt, 1963).

4.5 Factors and theories explaining M&A activity

As referred to in chapter 3, the overall evidence suggests that diverse macroeconomic variables are associated with the timing of M&A. The literature has developed several hypotheses about the relationship between M&A and business cycles, and/or financial markets. Some examples of literature, using aggregate or industry-based data, include Weston (1953), Markham (1955), R. L. Nelson (1959), Weston (1961), Gort (1969), Singh (1971), Steiner (1975), Beckenstein (1979), Melicher et al. (1983), Guerard (1985), Becketti (1986),

163

Golbe & White (1988), Weston et al. (1990), Blair & Schary (1993), and Mitchell & Mulherin (1996).

The M&A activity is procyclical, as generally it leads the business cycle slightly (see e.g. Golbe & White, 1988; R. L. Nelson, 1959; Steiner, 1975; Weston et al., 1990). According to Weston (1990), the activity is also approximately coincident with share price movements. Several authors find that share prices lead the M&A activity (e.g. R. L. Nelson, 1959), while others conversely conclude that M&A activity lags the stock market movements (e.g. Melicher et al., 1983).145 This divergence of findings may be partially explained by changes in markets trends, but also by the time spent between the beginning of the negotiations and the accomplishment of the deal. In the 1970’s, Halpern (1973) and Mandelker (1974) find this period to be on average about six months. Therefore Melicher et al. (1983), who evaluated share prices changes to precede M&A completed deals by one quarter, concluded that M&A negotiations lead share price movements by about one quarter.

In addition to macroeconomic factors, microeconomics also plays an in important role. For example, Lev & Mandelker (1972) discussed the microeconomic consequences of mergers, which generated a reply from Reid (1975). More recent literature argues that both endogenous and exogenous effects are present in M&A activity. This finding was made clear, for instance, at industry-level literature (see e.g. Blair & Schary, 1993; Cabral, 2000; Mitchell & Mulherin, 1996). Therefore, it seems that the development of a theory on M&A should encompass both features.

In terms of endogenous factors, time, seasonality and particular patterns within the M&A activity seem to contribute to partially justify the M&A occurrence behaviour. Effects from holidays and certain weekdays, weeks, months, or

145 According to Mueller (1980), in West Germany during the 1960s, M&A activity lagged behind share prices. However, in the 1970s, M&A activity tended to lead share prices and other aggregate measures, such as GDP and gross fixed investment.

164 quarters, often offer remarkable justifications for the occurrence of M&A deals (vid. e.g. Branch et al., 2001; Hirsch, 2004).

Among the exogenous factors related to M&A activity, the majority of literature suggests a relationship not only with stock market prices, but also with industrial production (e.g. Gort, 1969; Markham, 1955; Mitchell & Mulherin, 1996), and with interest rates (e.g. Beckenstein, 1979; Becketti, 1986; Golbe & White, 1988; Melicher et al., 1983; Weston et al., 1990). Extraordinary events, such as the case of the 11 September 2001 attacks on the USA, may also play an important role, while specific industry-deregulatory shocks may result in more long-lasting effects (vid. e.g. Andrade et al., 2001; Harford, 2005; Mitchell & Mulherin, 1996).

Both endogenous and exogenous factors to the M&A activity are discussed in more depth in the models development’ chapter of this thesis. In a different stream, but also playing a main role in the developing of the research hypotheses, is the awareness of the implications arising from the existence of accounting choice, and the effects resulting from changes in accounting regulation, particularly concerning the impacts on M&A activity that may have resulted as a consequence of the business combinations accounting changes.

As examined in previous chapters, a strand of literature has studied the market effects of the existence of two different accounting methods for business combinations: the purchase method and the pooling of interests method. It is important to recall that changes in accounting policy may produce economic effects and that accounting choice in M&A has also value implications.146 Several studies found that firms involved in M&A deals adopted an accounting method based upon certain financial and non-financial characteristics (e.g. M. L. Davis, 1990; Dunne, 1990). It has also been documented that managers

146 As discussed earlier, if accounting could be effectively neutral then the accounting standards would not produce economic effects. Therefore, if neutrality is not to be assumed, stewardship is required to deal with any significant economic consequences that may result from changes in accounting policy.

165 preferred pooling and that pooling firms willingly incurred significant costs to achieve the desired financial reporting outcome (Aboody et al., 2000; Ayers et al., 2002; Linsmeier et al., 1998; Lys & Vincent, 1995; Robinson & Shane, 1990; Walter, 1999; Weber, 2004). Despite the preference for pooling, however, empirical evidence supports market efficiency, which means that M&A is valued the same regardless the pooling versus purchase adoption (e.g. M. L. Davis, 1990; Hong et al., 1978 Lindenberg & Ross, 1999; Vincent, 1997).147 Nevertheless, existing literature also revealed that pooling resulted in mechanical effects on companies’ financial statements and on the analysis of the financial statements (Jennings et al., 1996; Vincent, 1997).

The replacement of the purchased goodwill amortisation method by impairment tests may also have had an impact on M&A activity. The research findings indicate that the market reacted negatively to the amortisation of goodwill by purchase firms (e.g. Ayers et al., 2002; Hopkins et al., 2000). Not surprisingly, several authors (e.g. Robinson & Shane, 1990) found that a higher bid premium, enhancing the size of the potential goodwill, increased the likelihood of pooling (Weston et al., 2004). Nevertheless, share prices should not decline significantly for companies with one-time impairment write-offs, unless they would become habitual (Hopkins et al., 2000).

Taking into consideration the anecdotal evidence and the literature findings examined before in this thesis, the first hypothesis to be tested is exhibited below in the null form:

Hypothesis 1 The FASB new pronouncements have had no impact on the number of M&A deals announced and completed.

The analysis of the testing results of this first hypothesis may provide evidence about any impacts on the M&A activity from the abolishment of pooling of

147 Nevertheless, according to Hopkins et al. (2000), analysts’ valuations were lowest when a company adopted purchase method and amortised goodwill.

166 interests and the replacement of purchased goodwill amortisation by impairment tests. Therefore, this hypothesis tests the appropriateness of FASB’s new rules, in the scope of the desired neutrality of the accounting standards. The testing also documents the influence of economic, financial and time factors to the pattern of M&A occurrence.

If significant results about impacts from the new standards on M&A activity are obtained, thereby preventing the rejection of this hypothesis, it would therefore be suggested that M&A activity is unrelated to FASB changes and is rather driven by financial, economic, time, or other factors. Conversely, rejection of hypothesis number one would suggest that FASB new pronouncements had produced a significant impact to M&A market participants and failed to minimise any possible economic effects. In this case, the other two possible hypotheses, called alternative hypotheses, are that the M&A activity benefited from the accounting changes, or that M&A activity did not benefit from the accounting changes.

4.6 Reasons explaining M&A abandonment

Being successful in M&A transactions seems far from easy (e.g. Kummer & Steger, 2008). Unsurprisingly, a substantial number of announced M&A deals are never completed (e.g. Muehlfeld et al., 2007; Ribeiro & Crowther, 2007a). For example, Pickering (1978) reports a 14% abandonment rate, while Muehlfeld et al. (2007) estimate it to be as high as 27%.148 In the period in between 2000 and 2002, the rate of deals announced but not completed in the USA was 20%.149 This fact is relevant, because M&A can be very expensive, and so can its abandonment, since firms need to allocate significant resources while planning and preparing a deal (see e.g. Weston et al., 2004). The literature has also detected a loss of value in the period between the time of announcement

148 Muehlfeld et al.’s percentage is not for aggregate data, but only for newspaper industry during the period 1981-2000. 149 Author’s estimation (source: Thomson Financial, 2006).

167 and withdrawal, particularly for target companies (see e.g. Bradley et al., 1983; Fabozzi et al., 1988).

It is important to stress that literature concerned with the study of M&A abandonment causes is scarce. The majority of the studies focus on the post- -M&A period analysis; only a few are concerned with the examination of the pre-completion phase and with M&A cancellations.150 Muehlfeld et al. (2007) point out a major difficulty related to this type of analysis:

“Decision-making processes at the pre-completion stage are largely unobservable to financial markets and difficult to capture based on accounting data.”.

Despite the non-existence of a global theory, existing literature provides some evidence to help explaining these occurrences. The explaining factors are mainly related with the type and way of concretisation of the M&A deal. Bidder and target firms’ characteristics and attitudes also play key roles. Furthermore, it is not easy to track the route of a M&A bid, as it may roll over a large period. Bradley et al. (1983) note that a potential M&A cannot be immediately classified as withdrawn if the first bid fails since there is the possibility of subsequent bids. In fact, a long period of time may pass before the M&A deal can be confirmed as withdrawn, cancelled, or unsuccessful.

Dodd (1980) emphasises that M&A bids and proposals are subject to discretional decision from the management. The target firm’s shareholders delegate the decision to the management, but hold the power to vote after their recommendations have been made following M&A proposals. Nevertheless, the management has the power to decline any friendly M&A proposal without presenting it to the shareholders. According to Davidson III et al. (2002), this power can be regarded as a safeguard for the firm, insuring that the M&A deal is adequate (Franks & Mayer, 1996), but, conversely, it can also be perceived as

150 Examples of literature concerned with pre-completion and M&A cancellations include Asquith (1983), Wong & O’Sullivan (2001), and Davidson III et al. (2002), as discussed in this thesis’ section.

168 an instrument of protection for the management, used with the purpose of avoiding the loss of their own positions in the target firm as a consequence of a successful takeover, at shareholders expense (Karpoff et al., 1996). Consequently, cancelled M&A deals often reflect an agency theory issue, where the interests of the management do not match with the interests of the shareholders (Davidson III et al., 2002).

Concerning the bidder’s attitude, Holl & Kyriazis (1996) point out that hostile takeovers are more likely to meet resistance from target firms. Negotiations that start friendly often result in disagreements. This makes transactions more costly and increases the likelihood of a bid cancellation. If a bid is considered friendly, one could expect it to be less dependent from negotiations to be successful, as it would be less susceptible to face resistance from the target firms’ management teams (Wong & O’Sullivan, 2001).

Wong & O’Sullivan (2001) also suggest that the method of payment may also help to explain the M&A abandonment phenomenon. Cash is easy to value and makes the bid more attractive to target firms’ management and shareholders. Consequently, its use increases the possibilities of M&A completion, since it reduces the prospects of disagreements between participants during the negotiations.

Finally, in what concerns the implications of regulation issues in business combinations accounting to the completion of M&A deals, it is important to bear in mind that FASB’s new M&A accounting standards have been the focus of much controversy, being arguably one of the most important statements published in recent years, as they transformed deeply the accounting for business combinations in the USA (Ribeiro & Crowther, 2005; Zeff, 2002). As referred to earlier in this thesis, in an unprecedented effort, between 1996 and June 2001, the FASB (2001c) had to deal with some fierce opposition and lobbying, and had therefore to be very diligent in order to adopt the new regulation, as proved by its intense activity, which included issuing four documents for public comment, holding over sixty public meetings, conducting

169 public hearings and visits, and analysing and discussing more than five hundred comment letters.151 Indeed, although accounting practitioners and academicians in general supported purchase as the single method for M&A accounting, many firms disagreed, vigorously opposing the pooling elimination.152 It is therefore reasonable to question whether the challenged, disputed, and controversial changes in business combinations accounting constrained the M&A activity, resulting in an increase of the number of M&A deals withdrawn.

Considering the existing literature and evidence, and the objectives of the present study, it becomes possible to test another general hypothesis, stated in the null form:

Hypothesis 2 The FASB new pronouncements have had no impact on the number of announced M&A deals withdrawn.

If M&A deals, which were previously intended and structured for pooling of interests, can no longer qualify for pooling following this business combinations method elimination, one could expect an increase on the number of M&A withdrawn, as a consequence of FASB’s new pronouncements. This is the main suggestion underlying the hypothesis stated above. Nevertheless, despite making no explicit mention, hypothesis two goes beyond the scarce findings and evidence available in existing literature as it implicitly includes economic, financial, and time variables as potential explaining factors of M&A deals cancellations, i.e., the endogenous and exogenous factors referred during the development of hypothesis one, may also contribute to justify the withdrawal of

151 Including an Exposure Draft (1999a) and a Revised Exposure Draft (2001b). 152 As discussed earlier, for example, Dennis Powell from Cisco Systems, warned about the potential negative effects on the USA economy (Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000), while Jim Barksdale, former CEO of Netscape, declared: “AOL/Netscape merger would not have occurred if pooling had not been an option” (Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, 2000).

170

M&A deals, as to be discussed more in depth in the models development’ chapter.153

Similarly to hypothesis one, this hypothesis will also test the appropriateness of FASB changes in the scope of the desired neutrality of the accounting standard- -setting policy.

Not rejecting hypothesis number two would suggest that the phenomenon of withdrawn M&A deals is unrelated to the FASB changes and is rather explained by economic, financial, business conditions, or other factors. In opposition, the rejection of the hypothesis would imply the acceptance of the alternative hypothesis that FASB’s new pronouncements led to an increase of M&A deals cancellations. Furthermore, even if seemingly not so reasonable from a pure theoretical point of view, in case of rejection one could also hypothetically admit another alternative hypothesis: that FASB’s new pronouncements resulted in a decrease of M&A deals cancellations.

4.7 Conclusions

The main purpose of the research has been discussed in this chapter: the investigation of the existence of any economic effects resulting from changes in accounting policy’s on M&A activity. A set of hypotheses has been placed to allow assessing about the impact of such changes. The hypotheses have been formulated in a general form. Although it is possible to extract sub-hypotheses from the main hypotheses, it is not intended to do so, as it is not a purpose of this research to examine particular characteristics related to M&A activity, and to the economic-financial environment, but instead to focus on the possible impacts from the regulation changes.

153 An example of parallelism between common determinants for M&A completion and M&A cancelations follows: while a decrease of interest rates is most likely to contribute to an increase of the number of M&A deals completed, a decrease is more likely to result in an increase of M&A deals withdrawn.

171

Furthermore, evidence on possible effects from managerial decisions about M&A, and on financial reporting, is to be presented and discussed in chapter 5, which follows next. However, such evidence, which has been collected in the early stage of this research, will not be tested having research hypothesis status, as it will serve mostly to enrich the discussion of the results obtained from the main hypotheses testing. Despite this scope limitation, one can argue that this investigation is threefold, since it studies the possible effects of business combination accounting changes on M&A activity, managerial decision-making, and on financial reporting.

Finally, the methodology adopted to test the research hypotheses is shown in chapter 7, together with the discussion of the results obtained from the tests performed; while the justification of the selection of the samples for analysis, and the rationale for its different sets of aggregation, is made in chapter 6.

172

Chapter 5 Survey and Financial Reporting Analysis

5.1 Introduction

In previous chapters, a qualitative analysis has been undertaken, by means of literature review. Such qualitative analysis discussed whether the changes in business combinations accounting occurred in 2001 may have produced significant effects in M&A activity in the USA. Indeed, FASB’s proposals generated widespread concern and criticism from the business community, which was made very clear by representatives of several industries, such as IT, financials, industrials, or healthcare. Moreover, as examined quantitatively in the preceding chapter, it seems obvious that there may have been some very short-term impacts on M&A activity, at the very least, being this particularly evidenced in the weekend when the new accounting standards were made effective.

Following the doubts cast by the performance of both qualitative and quantitative analyses, it seems therefore relevant to investigate whether the accounting changes brought any impact on M&A deals completion from a companies’ perspective. Accordingly, a questionnaire was addressed to the top management with financial reporting responsibilities from some of the 500 most important companies in the USA. Additionally, as changes in GAAP may not only produce economic consequences, but also affect significantly the financial reporting, the impact from the new M&A accounting rules on annual reports was measured and examined as well.

173

The evidence revealed in this chapter, the survey and the annual reports’ analysis, is therefore to be triangulated with the results obtained from the research hypotheses testing, as to be shown in following chapters. Finally, the discussion of the implications of this findings’ triangulation is to be continued in the final chapters of this thesis.

5.2 Survey on companies’ reactions to the new M&A accounting rules

As discussed in earlier chapters, the strong reaction from significant segments of the corporate world to FASB’s proposals on business combinations accounting suggested that the adoption of new GAAP could constrain the completion of M&A deals. Even after FASB compromised, agreeing to revise some of the initial exposure draft proposals, the opposition to pooling of interests discontinuation continued to be expressed publicly by pooling supporters.

Therefore, as an initial step of this thesis’ research, an exploratory survey was addressed to the companies included in S&P 500 index, by means of a questionnaire, in order to assess the managers’ perception about the relevance of the business combinations accounting changes to the M&A activity in the USA. The list of the 500 companies composing the S&P 500 index as at the end of 2004 is shown in Appendix A.154

The adoption of S&P 500 index is justified by its representativeness of the corporate environment in the USA. As shown in Appendix A, almost all S&P

154 The use of companies composing the S&P 500 index by the end of 2004 is justified by the purpose of having the maximum of active companies that could participate in the survey, which was launched in early 2005. The economic downturn and the wave of corporate scandals in the early 2000’s led to many corporate failures, by means of bankruptcies or M&A deals. Concurrently, some companies were delisted, or replaced in the S&P 500 constituents list. Therefore, the use of the companies listed in S&P 500 index during the first years of the new millennium would result in a smaller response rate, simply because some companies’ existence was transformed or ceased.

174

500 companies have their headquarters in the USA, apart from four companies which are located in small Caribbean territories, but with USA and global-based operations.155 Moreover, S&P 500 companies are highly publicised and tracked by data providers, being much easier to find all sort of institutional and financial information that is made available in several informational platforms.

Not all S&P companies have been necessarily involved in M&A deals, and therefore it could be argued that they should not be considered. Nevertheless, as information about companies involved in M&A deals was missing, it was decided to include every S&P 500 company in the survey in order to: i) increase the odds of a higher number of responses, thereby increasing the possibility of collecting a sizeable sample; ii) enhance data triangulation with annual reports data analysis; iii) include a high diversity of companies and different industries; and iv) ensure getting feedback not only from companies involved in the past in M&A deals, but also from others not involved but potentially intending to be.

There are also companies which are not supposed to be involved in M&A due to the nature of their business, as it is the case of Fannie Mae and Freddie Mac. For these companies, and for other that have never been involved in M&A deals, the absence of feedback could be expected. Nevertheless, any contribution from these companies could be taken into consideration, as it could reveal some kind of interest by the possible respondents on the topic.

Earlier, in chapter 4, some literature concerning studies using M&A surveys was examined and may also be referred to whenever appropriate in this chapter, together with other references, as follows. The mail questionnaire format was chosen, since the target population was based overseas. Moreover, mail questionnaires are more frequently used for social research than either telephone or face-to-face interviews (Dillman, 1991). The questionnaire was designed following basic standards for mail surveys (vid. e.g. de Vaus, 2002; Dillman, 1991), as little guidance about international mail surveys can be found in

155 The headquarters countries shown in Appendix A were mostly taken from EDGAR Online Pro database during 2005. News Corporation was registered in Australia until 2004.

175 literature (Dawson & Dickinson, 1988; Harzing, 1997; Jobber & Saunders, 1988), even if this shortage is being reduced in recent years (vid. e.g. Harzing, 2006). In any case, this thesis’ survey is not truly international, in the sense that it does not try to capture international differences; therefore the most relevant contributes from this strand of literature are related with respondents’ approaching procedures and response rates, as shown later in this thesis’ section.

Due to the nature and to the complexity of the subject matter, the questionnaire was addressed to individuals with superior responsibilities in the company.156 This procedure finds parallels in other studies, such as in the one conducted by the Association for Financial Professionals (2000), which surveyed senior level financial professionals. The questionnaire was designed to be as simple and concise as possible in order to increase the odds of obtaining a reasonable response rate from people with limited time and availability (vid. e.g. Frary, 1996). Accordingly, the questionnaire was composed of 5 questions only, all of which could be answered in a very simple way, by making a cross in the appropriate field. As shown in Appendix B, the questionnaire occupied a single page.157

The questions were constructed primarily on the basis of the expectations and concerns expressed on questionnaires prepared by the Association for Financial Professionals (2000), Moehrle et al. (2001), and on the feedback received by the FASB (Financial Accounting Standards Board, 2001a, 2001e). Although the questions were formulated primarily as closed, i.e., they could be answered with either a single word or a symbol, four of them were simultaneously open-ended, optionally allowing the respondents to express concise comments at their own

156 The personalisation of the questionnaire was possible following the selection of top managing officers, made through consultation of information available in annual reports and corporate webpages. 157 The version shown in the appendix is not scaled to the original version, as the answers’ fields were downsized in order that the questionnaire could fit in a single page according to this thesis’ definitions.

176 discretion. The objective of using close-ended questions was not only to avoid variation in respondents’ willingness to respond in writing (Brace, 1994; Frary, 1996), but also to allow the use of statistical inference, thereby sophisticating the survey results’ quantitative analysis. The objective of the use of optional open-ended questions was to allow also some qualitative analysis.

The survey was conducted through postal questionnaires. A covering letter stressed that the questionnaires would be treated as strictly confidential, in order to increase the receptiveness to feedback, even knowing this to not necessarily to increase the respondent’s willingness to respond (vid. e.g. Singer et al., 1992). Nevertheless, as M&A deals and business combinations accounting are very sensitive topics, it was therefore necessary to reassure confidence to the potential respondents as much as possible.

A 10% response rate was set as a desirable target, knowing the response rate could be low in mail questionnaires, while also being aware that a low response rate does not necessarily mean that the data collected is biased, as long as the survey targets an homogenous population, as it is the case in this thesis’ survey (Leslie, 1972). Furthermore, Visser et al. (1996) demonstrated that surveys with lower response rates, near 20%, provided more accurate measurements than surveys with higher response rates of around 60 to 70%. Additionally, Holbrook et al. (2007), while examining the results obtained for 81 surveys with response rates varying from 5 to 54 percent, found that surveys with much lower response rates were nevertheless only minimally less accurate.

A first wave of 500 questionnaires was posted in early 2005. As one would expect, the immediate response rate was low: 3.8%, perhaps due to the fact that the respondents were based overseas. Jobber & Saunders (1988) tested the hypothesis that managers from the USA are more likely to respond than those from the UK. Nevertheless, they found that responses from managers in the USA were no higher than from counterparts in the UK, suggesting that foreign source effects do not necessarily raise response rates (Jobber & Saunders, 1988). This finding does not corroborate literature suggesting that foreign surveys

177 generate higher response rates than domestic ones (vid. e.g. Ayal & Hornik, 1986). Furthermore, Jobber & Saunders (1988) found evidence contrary to their own expectations, and suggesting the opposite, i.e., that domestic surveys generate higher response rates than foreign surveys.

As the initial response rate was below target, a polite request was prepared which resulted in a second wave of 481 questionnaires which were posted a few weeks later. The response rate for this insistence was of 3.4%. As the sample collected was not considered large enough, a more direct approach was then followed consisting in electronic insistences (vid. e.g. Dillman et al., 2009). Accordingly, e-mails from the most appropriate individuals and corporate areas were collected from companies’ webpages and also from corporate databases. In some cases, it was not possible to use e-mail as a way of communication, as it was found to be necessary to submit specific forms using the companies’ webpages. In rare cases it has been necessary to call companies in order to get contact e-mails. Not uncommonly, it has been necessary to send several e-mails to different people in different positions in the same company, in order to find the right person to whom to send the electronic questionnaire.

Overall, replies from the initial questionnaire and insistences were received from a total of 56 respondents, corresponding to a global response rate of 11.2%. However, some respondents declined to answer the questions, while others felt that the questionnaire was not applicable to the scope of their organisations, as M&A deals were non-existent or considered to be irrelevant to the company. Consequently the effective response rate was of 10.4%. The final sample is composed by 52 questionnaires sent and received throughout 2005.

The questionnaires were sent to companies from 10 different industries, but no questionnaires were received by companies from the three following industries: energy, telecommunication services, and consumer staples. Despite this, the spectrum of corporate sectors that comprise the sample is quite diversified and well-weighted, as exhibited in Fig. 5.1 below. Financials, with almost one-third

178 of the replies, is the most represented sector, followed by industrials and utilities. Information Technology Health Care 7.7% Financials 9.6% 30.8%

Materials 9.6%

Consumer Discretionary Industrials Utilities 12% 17.3% 13.5%

Fig. 5.1 Information on the respondents’ corporate sector Percentages do not total 100.0 due to rounding.

One of the main concerns of the survey was to get feedback from people with effective responsibilities over accounting and M&A departments. Therefore, the questionnaires were addressed individually to the person on the company’s board perceived to be more qualified to participate in the study. As shown in the Fig. 5.2 below, this purpose was achieved: two-thirds of the respondents were controllers, and the remaining respondents’ occupied relevant positions in the company.

179

Treasurer CFO Principal 7.7% 1.9% Accounting Officer 9.6%

M&A/

/Corporate Controller Development 67.3% 13%

Fig. 5.2 Information on the respondents’ professional background Percentages do not total 100.0 due to rounding.

As far as the author is aware, only 15.4% of the responding persons were different from those to whom the questionnaires have been addressed.158 Finally, there was a case of a company that was not involved in any M&A deal for decades, but nevertheless replied to the questionnaire. The evidence provided by that company was fully considered and treated evenly, not only because it was a single situation, but also because of the reasons presented before in this chapter.

The first two questions of the questionnaire aimed to identify the possible effects of the new accounting standards upon the completion of a M&A by a company. As suggested by Fig. 5.3, the effects appear to have been minimal.

158 The questionnaire’ cover letter solicited the identification of the person completing the questionnaire, in case of being different from the person to whom the questionnaire had been addressed.

180

Contributed 1

0,5

SFAS 142 SFAS 141 0

No effect Constrained

Fig. 5.3 Impact of the new accounting standards on the completion of M&A deals

With regard to SFAS 141, 92.3% of the respondents believed that it did not produce any effect, while the remaining 7.6% considered that pooling discontinuation constrained the M&A activity. Most respondents who pinpointed the existence of effects were from the financial and IT industries, 3/4 of the total, more precisely. Further detailed information about questions one and two can be found in Appendix C.

In terms of qualitative analysis, a respondent from the information technology industry justified the negative impact of SFAS 141 on the completion of M&A deals through the following comment:159

“Constrained. Near term earnings impact of amortizations limit the M&A options, despite the fact economic differences of pooling/purchase can be disputed.”.

None of the respondents to this thesis’ questionnaire mentioned any positive effect on M&A activity from SFAS 141 adoption.

159 Every respondents’ comment directly related to this research questionnaire is quoted in this section of the thesis.

181

As regards the effects of SFAS 142, which replaced goodwill amortisation by impairment tests, the range of the answers was wider, although the majority also believed it has neutral effects: 76.9%. About 9.6% of the respondents believed the SFAS 142 changes contributed to the M&A activity. Once again, 3/4 of the respondents that pinpointed the existence of effects, were from financials and IT industries (see Appendix C).

In terms of further comments received for question two, one respondent from utilities industry, safeguarding not being the case for the company, argued that:

“Contributed. Not specific to our company, but in general it is easier to make the deal accretive if there is no goodwill amortization”.

However, a higher number of respondents, 13.4%, supported exactly the opposite. One respondent from the financial industry argued that the changes in SFAS 142 were negative to the M&A activity by stating:

“Constrained. The organization has become much more sensitive to amount of goodwill a transaction would add to balance sheet. Relationship of goodwill to equity has become a very important step”, while a respondent from a different industry, information technology, pinpointed that

“Constrained. Similar impact with more valuation allocated to amortizable assets”.

These worries, with goodwill recognition and its impact on the financial reporting, corroborate the concerns expressed in evidence collected by Moehrle et al. (2001: 119):

182

“As an initial step in this study, we surveyed Chief Financial Officers (CFOs) of each of the firms represented in the treatment and control samples. One questionnaire item requested reasons that the CFOs of pooling firms chose to structure the deal as a pooling. Of the 14 pooling CFOs that responded, 10 (71 percent) indicated that pooling was selected to avoid the drag on earnings. Likewise, the CFOs of purchasing firms were asked why the merger was structured as a purchase. Their responses generally reflected economic and cash flow reasons. In addition, pooling CFOs were 3 times more likely (84 percent vs. 25 percent) to indicate a belief that the market places undue weight on reported earnings for valuation purposes. Thus, anecdotal and survey evidence supports the claim that the preference of many managers for pooling accounting treatment results from a desire to avoid the earnings decrease created by purchase accounting.”.

In conclusion, despite a great majority of answers suggesting the neutrality of accounting changes to the M&A activity, there are however dissenting answers. Moreover, some of which were supported by sound argumentation, therefore deserving to be considered.

The third and fourth questions inquired whether companies had withdrawn any planned or announced M&A due to accounting regulations. All the participants declared that they did not face this situation. Some rightly assumed the prevalence of strategy over financial reporting, admitting to cope with accounting issues:

“To my knowledge, our organization has not withdrawn any planned or assumed M&A deals due to accounting regulations. Our M&A policy/strategy is driven by strategic considerations and once a target is identified, we work within the current accounting guidelines to make the deal most attractive from a GAAP perspective”; while others implicitly conceded that accounting could influence strategic M&A decisions:

183

“We try not to let accounting guidelines drive our strategic M&A discussions”; and that ultimately it can obstruct the completion of a M&A deal:

“Never. But limit opportunities that made it through our”.

When queried as to whether the corporation brought forward any M&A deal to try to qualify it for pooling of interests before its elimination, avoiding revised purchase accounting method, effective after July 1, 2001, all respondents asserted that it did not happen, with answers from a straight:

“Never”, or,

“Not that I'm aware of”, to another answer referring that the situation was somewhat not applicable to the company as

“Our preference has always been the purchase method”.

There were, however, some elaborated comments:

“Before 7/1/01, the company completed major acquisitions (...) which were pooling of interests. The new pronouncements clearly has increased asset carrying values, but the oil & gas industry had not had goodwill before 7/1/01 so a fair amount of the asset step up is goodwill now, and has not stopped us from doing future deals.”, and, finally, an interesting quote from a company from the materials industry:

184

“None. Even through pooling of interests was enormously preferable to the silly inflated-tech purchase accounting. In any case its the economics that count not the accounting.”.

Despite this final comment being made in the fourth question, it relates more to question one. This curious comment contradicts the single comment received for question one, made by a company from the information technology industry, which argued that pooling elimination would limit M&A options. This, perhaps bitter, comment also reveals a certain degree of resentment from a company belonging to the “old” economy against the excessive focus on the hi-tech industry, which by then was catalogued to belong to a kind of fashionable “new e-economy”. One thing seems to be evident: the regulatory needs vary among industries, which was also reflected in the comments received.

Finally, question 5 was prepared to assess how the corporations’ administrators would classify the relevance and impact of the new accounting rules to the M&A decision-making and overall activity. Only a single comment has been received from an industrial company:

“No real impact due to accounting rule changes. M&A deals are evaluated on IRR basis and NPV which focuses on cash flows.”.

As shown in Fig. 5.4, the majority of the respondents believed both standards had a minor degree of importance.

185

No. of Companies

40 SFAS 141

30 SFAS 142

Low Medium High

Fig. 5.4 Relevance and impact of the new accounting standards on the M&A decision-making and overall activity

The reduced level of importance is particularly clear for SFAS 141, as only one respondent, from financials industry, attributed a medium importance to the accounting change that resulted in pooling of interests discontinuation. As for SFAS 142, 5 respondents classified the new standard as important, and 3 other found it to have a medium importance. Most of the respondents that pinpointed some degree of importance for SFAS 142 were from financials industry. This evidence suggests that corporate managers perceived the new accounting treatment to purchased goodwill to be more relevant than the elimination of pooling of interests. crosstabulations and tests of independence for the survey results: Since representatives from the IT and the banking industries were among some of the most active opponents to the business combinations accounting changes, as demonstrated in the literature review made, and as shown in the results of the survey, some statistical testing is justified, in order to try to access whether such evidence has some substance, and is not simply the result of some sparse coincidence.

The testing methodology involved crosstabulations and related tests of independence. The crosstabulation procedure summarises two columns of

186 attribute data, by constructing a two-way table showing the frequency of occurrence of all unique pairs of values in the two columns, i.e. representing the pairs of variables (StatPoint Inc., 2005). While several statistics can be constructed in order to quantify the degree of association between the columns of data, tests can be run in order to determine whether or not there is a statistically significant dependence between the values from the different columns (StatPoint Inc., 2005). The use of crosstabulations offers several advantages, such as: ease of visualisation and interpretation, allowance of using a large number of statistics, and construct procedure simplicity (vid. e.g. G. M. Phillips, 1995).

In terms of tests of independence, a common question asked when data is organised in two-way tables is whether or not the row and column values are independent (StatPoint Inc., 2005). In the case of this thesis’ questionnaire, feedback from questions one and two, effects of SFAS 141 and SFAS 142 respectively, were crosstabulated with IT and financial industries, being the frequencies displayed in tabular form, as shown in Appendix C. Every answer referring that the accounting changes contributed or constrained the M&A activity has been considered as “effect” versus the remaining “no effect” answers. If rows and columns are independent, then the fact that a response is placed in a particular row does not affect the probability of its falling in a given column. In the case of this research questionnaire, for a given crosscorrelation independence, it would be implied that a pinpointed accounting standard effect had no relationship to the fact that the respondent belonged to the financials and/or IT industries.

In terms of tests available, there are many statistics related to crosstabulations, such as Chi-squared, or the Lambda coefficient. Some statistics are suitable to be used when variables are measured at the nominal level, such as the case of Cramer’s V, Contingency coefficient, and Phi coefficient; while others, like Gamma test, or Kendall’s tau “b” and “c”, are to be used when variables are measured at the ordinal level, which is not the case of the variables constructed from questions one and two.

187

Despite the large amount of statistics available, economists appear to favour Pearson Chi-squared statistic, i.e. the most common Chi-squared test form (Gaughan & Thomas Hodson Jr., 1993; Mansfield, 1986; G. M. Phillips, 1995; Piette, 1992). This preference is unsurprising, as Chi-squared test is overall the most common test for independence in two-way tables (StatPoint Inc., 2005). 2 Chi-squared test, or x , compares the observed and expected frequencies, Oij and

Eij respectively, for i rows and j columns, by computing:

rc()OE− 2 χ 2 = ∑∑ ij ij ij==11 Eij

In a typical sample differentiation analysis, the crosstabulation analysis is made by using Chi-squared to test the null hypothesis of independence between rows and columns (e.g. Mansfield, 1986; G. M. Phillips, 1995). If the test shows no significant differences, usually measured by a p-value greater than the 5% critical value, then one would not reject the null hypothesis of no differentiation (G. M. Phillips, 1995). Conversely, as Table C.3 and Table C.4 exemplify, when Chi-squared p-value is smaller than 5%, then one can reject the hypothesis of no differentiation, meaning in this case that the answers from a particular industry are linked to a particular pattern of feedback.

There is, however, an issue with Chi-squared: its use may be problematic whenever any cell value has a joint frequency of less than five (Fienberg, 2007), which is the case of all tables, except Table C.6. Therefore, other tests have also been used, namely Pearson’s R, which test the strength of relationship of the crosstabulations with values ranging from -1, i.e. perfect negative correlation, to +1, i.e. perfect positive correlation, being zero values an indicator of association absence. Also shown in Appendix C’ tables are the Contingency coefficient, and the Cramer’s V statistic, which are more suitable to test the strength of association of the cross tabulations when the variables are measured at the nominal level. Their values range from 0, i.e. no relationship, to 1, i.e. the maximum rank possible of association.

188

In terms of overall results, it is possible to conclude that the FASB’s changes were unrelated with to the IT and the financial industries when taken together. There has also been no evidence found suggesting that the effects from the new accounting standards were significant to the respondents from the financial industry. However, the hypotheses that SFAS 141 and SFAS 142 were independent from the IT industry could be rejected at the 95% confidence level, with a certain degree of association between variables, as shown by the Contingency Coefficient, Pearson’s R, and Cramer’s V tests. This statistical finding needs however to be regarded with some precaution, as only four IT companies answered the questionnaire. In opposition, as 16 financial companies participated in the survey, it seems clearer to admit that the associations of the financial industry’s feedback with an accounting standard effect are not strong enough, or at least are not statistically significant.

5.3 Analysis of annual reports

Overall, from a managerial perspective, the results shown previously suggest that the new business combinations accounting standards did not result in significant effects on the M&A activity, despite some indications of a certain degree of impact on IT, and also on the financial industry. Nevertheless, other kind of effects may have been registered, namely on financial reporting. Accordingly, this section examines whether the new business combinations accounting rules had any significant influence on the financial reporting of companies involved in M&A deals. This is arguably a relevant issue because, as examined in earlier chapters, the changes in business combinations accounting generated much controversy as they often lead to significant changes in financial reporting.

Together with the survey, the analysis of financial reporting is indeed important, as it may help to understand how reasonable were the allegations from some industries, such as IT and financials, that the M&A accounting rules

189 changes would result in economic effects, by negatively impacting the financial reporting of companies interested to be involved in M&A deals, therefore inhibiting companies and resulting in a limitation on the occurrence of M&A deals, with damaging effects for the economy of the USA, as discussed earlier.

5.3.1 Methodology of analysis

The survey methodology has been used in the previous section of this chapter, but in this section the methodology is based on the analysis of financial reports instead. Through the information available on the annual reports of S&P 500 firms, particularly in statements on accounting changes, mostly in the financial statements notes, the effects on financial reporting due to the new business combinations accounting rules were examined. The financial reporting analysis also allowed estimating the impact of the new accounting standards on corporate earnings.

The analysis of the effects on financial reporting is not intended to formally test any hypothesis. The analysis has a simple proposition, which is to verify whether the changes in business combinations accounting resulted in significant impacts on the financial reporting. This statement could be regarded as a testable hypothesis. However, in the present research, not only has it not been taken as a hypothesis, but it has also not been tested by means of any statistical analysis. Therefore, this proposition will only be evaluated by means of qualitative and quantitative analysis of financial reporting data. A part of the quantitative analysis will be cross-sectional.

5.3.2 Data sources

The sample for analysis, shown later, is the result of the congregation of the data collected from financial reports of companies that completed M&A deals in recent years, and that have reported business combinations accounting changes

190 following the adoption of the new FASB’s standards. As for the survey, the companies composing the S&P 500 index, effective after the close of 31 December 2004, were the centrepiece of the study, and therefore data has been collected from these companies’ financial reports. S&P 500 index includes firms with large capitalisation values, listed on both New York Stock Exchange (NYSE) and NASDAQ. The fact that these companies were listed enhanced the odds of involvement in M&A deals, as exchange markets ease the concretisation of M&A deals. Nevertheless, as discussed earlier, this fact does not warrant that most companies were neither involved in M&A deals nor affected by the new M&A accounting rules. The list of the companies for which the annual reports were examined is shown in Appendix A.

In the USA, the Form 10-K is an annual filing, which provides a comprehensive overview of the company for the period of a fiscal year. 10-K forms are often included in annual reports, although they are formally prepared to comply with SEC’s, and the Securities Exchange Act requirements, while annual reports are primarily focused in investors. Since companies have to report material changes as a result of the adoption of new accounting standards, it is therefore possible to collect data directly from annual reports and 10-K forms.160 The data for this thesis was collected primarily from the 10-K forms.161 The SEC Filings and Forms (EDGAR), and the EDGAR Online Pro database, were the main data sources used. Data from annual reports was also examined in order to collect additional information about some companies. The data sources for annual

160 In this thesis, “annual report”, “Form 10-K”, and “annual filling” are interchangeably used. 161 In some cases, the Form 10-K was not available and the Form 10-K405 was examined instead. Like the Form 10-K, the Form 10-K405 is also a SEC filing, but indicates a file violation resulting from a lack of disclosure of insider trading activities from the reporting company. The identification of the form as 10-K405, versus ordinary 10-K, is made by a company’s officer or director, and not by SEC officials. Unsurprisingly, this managerial discretion resulted in inconsistent designations adopted by different companies. Consequently, the SEC’s Branch of Public Reference discontinued the requirement to designate a filing as a Form 10-K405, effective after 2002, being 10-K405 forms no longer accepted by the SEC filings system.

191 reports were The Annual Reports Service, from Barron’s, and several S&P 500 companies’ websites.

5.3.3 Data collection

Since both SFAS 141 and SFAS 142 became effective during 2001, annual reports and 10-K forms have been examined since fiscal year 2001. As exhibited in Appendix A, most companies have a fiscal year-end in 31 December, or by the end of the year, and therefore the changes in business combinations accounting occurred in 2001 were likely to have been disclosed immediately, i.e., from 2002’s filings onwards. However, as discussed later in this section, filings up to year 2004 could keep referring to 2001’s changes in business combinations accounting. new business combinations accounting standards disclosures: Both new business combinations accounting standards contain detailed provisions concerning disclosures on business combinations and on accounting changes, as shown in paragraphs 51 to 58 of SFAS 141 (Financial Accounting Standards Board, 2001a), and in paragraphs 44 to 47 of SFAS 142 (Financial Accounting Standards Board, 2001e). They are mostly to be presented in the notes to the financial statements. SFAS 142 also provides further guidance on disclosures in its Appendix C, by means of illustrations. Much of the impacts measured and reported by the sampled companies were based in such guidance.

An example of financial reporting disclosure on business combinations’ new accounting follows for 3M Company. For the year ending in 31 December 2001, 3M Company, registered as Minnesota Mining and Manufacturing Company, filed the Form 10-K405 in 11 March 2002, as required by the Securities Exchange Act of 1934. By standard, every reporting company devoted a section, or specific paragraphs, announcing the enforcement of the new business combinations pronouncements. In the first year of SFAS 141 adoption, 3M Company reported (Item 7, 2002):

192

“In June 2001, the Financial Accounting Standards Board issued SFAS No. 141, Business Combinations. SFAS No. 141 applies to all business combinations with a closing date after June 30, 2001. The most significant changes made by SFAS No. 141 are: 1) requiring that the purchase method of accounting be used for all business combinations initiated after June 30, 2001, and 2) establishing specific criteria for the recognition of intangible assets separately from goodwill.”.

This description configures a commonplace one, as it illustrates the way companies reported the generic changes in business combinations accounting. Overall, companies reported the existence of the new business combinations standard, SFAS 141, but no effects were to be disclosed, as pooling of interest was simply no longer an option. The adoption of SFAS 141 was therefore absolutely neutral from a financial point of view. When referring to both SFAS 141 and SFAS 142, 3M Company states that (Notes to Consolidated Financial Statements, 3M Company, 2002):

“These standards permit only prospective application of the new accounting; accordingly, adoption of these standards will not affect previously reported 3M financial information.”.

Since pooling of interests had been discontinued, no prospective application was therefore possible for SFAS 141. Companies simply informed whether they used pooling method before its elimination. However, a very different scenario was set for SFAS 142.

Like for SFAS 141, companies produced a similar standard description for SFAS 142 (Item 7, 3M Company, 2002):

“The Financial Accounting Standards Board recently issued Statement No. 142, Goodwill and Other Intangible Assets, which will be adopted by the company effective January 1, 2002. Goodwill and intangible assets acquired after June 30, 2001, are subject immediately to the non-amortization and amortization provisions of this statement, while existing goodwill and other

193

indefinite-lived assets will no longer be amortized beginning January 1, 2002. Goodwill will be subject to an impairment test at least annually”.162

However, a more detailed examination on the accounting changes resulting from SFAS 142 adoption is provided in the notes, as shown in the two paragraphs quoted below (Notes to Consolidated Financial Statements, 3M Company, 2002):

“SFAS No. 142 primarily addresses the accounting for acquired goodwill and intangible assets (i.e., the post-acquisition accounting). The provisions of SFAS No. 142 will be effective for fiscal years beginning after December 15, 2001. The most significant changes made by SFAS No. 142 are: 1) goodwill and indefinite-lived intangible assets will no longer be amortized; 2) goodwill and indefinite-lived intangible assets will be tested for impairment at least annually (…); and 3) the amortization period of intangible assets with finite lives will no longer be limited to 40 years.”.

Indeed, further examination of 3M’s Form 10-K405 reveals that the focus of business combinations accounting changes is on SFAS 142, as it is clearly assumed that:

“The principal effect of SFAS No. 142 will be the elimination of goodwill amortization. Amortization of goodwill and indefinite- -lived intangible assets in 2001 was $67 million (net income impact of $51 million, or 12 cents per diluted share)”.

The paragraph above identifies and illustrates the reported impact that is subject to analysis in this section of the thesis. Since the effects from SFAS 142’s adoption are to be measured by the companies, no computation is needed. This is an advantage for the researcher, as the risk of a biased data handling is

162 According to SFAS 142 (paragraph 43, 2001a), goodwill is the excess of cost of an acquired entity over the net amounts assigned to assets acquired and liabilities assumed in a business combination.

194 avoided. Any possible computation flaws are therefore responsibility of the reporting companies.

Like most of its peers, 3M Company was amortising goodwill in wide periods, including the maximum amortisation ceiling. As referred to by 3M Company in its annual report notes, in the first quarter of 2001 three notable business combinations were completed using purchase method. The purchased intangible assets, including goodwill, were being amortised on a straight-line basis over several periods, ranging from 4 to 40 years (Notes to Consolidated Financial Statements, 3M Company, 2002). As referred in the last quoted paragraph, if SFAS 142 would be adopted in 2001 the relief in losses as a consequence of the nonamortisation of purchased goodwill would be of $51 million alone for 3M Company. This would result in a significant increase of earnings per share in 2001: from $3.58 to $3.70, a net impact of 12 cents. Although 12 cents in $3.58 stand for only a 3.35% increase, evidence collected in this research, and to be examined later in this section, suggests the existence of significant higher possible impacts for most companies in the scope of the adoption of the new business combinations accounting standards.

The positive impact of the elimination of amortisation of acquired goodwill and other intangible assets on earnings was certainly mitigated by the recognition of impairment losses, particularly from companies holding higher figures of purchased goodwill in their balance sheets, also under SFAS 142.163 However,

163 Under SFAS 141, business combinations companies are required to estimate the fair value of acquired intangible assets in the following manner: first, intangible assets must be categorised by type, such as customer lists, trademarks, patents, software, intellectual property, etc; second, intangible assets with an identifiable remaining useful life must be separated from those with an indefinite useful life. The latter is then classified as goodwill and must be subject to a two-step test for impairment under FASB 142, which companies were required to adopt by January 1, 2002. However, if goodwill and other intangible assets acquired in a transaction for which the acquisition date is after June 30, 2001, but before the date of fully adoption of SFAS 142, these assets are to be reviewed for impairment in accordance with APB Opinion 17, or with SFAS 121, as appropriate, until the date that SFAS 142 is applied in its entirety (Paragraph 51, Financial Accounting Standards Board, 2001e).

195 from the analysis of the data collected from annual filings for the period 2002- -2003, it has been found that most S&P 500 companies did not recognise impairment charges following SFAS 142 adoption. In the fiscal years 2002 and 2003, at least 126 and 40 companies, respectively, disclosed some type of impaired purchased goodwill and other intangible assets than goodwill under SFAS 142.164 In the universe of S&P 500 index companies, these impairing companies figures would represent only around 25% and 8% in total, for 2002 and 2003, respectively. As to be shown later in this chapter, considering that 476 companies referred SFAS 142 adoption, these impairment figures would increase slightly only, to 26.4% and 8.4%, respectively.

Despite perhaps not being a significant number of S&P 500 impairing companies, there is, however, enough evidence suggesting that impairment charges were very significant in the first year of the new FASB’s standard adoption. Also, according to the evidence collected in the present research for

Under SFAS 142, the two steps of the goodwill impairment test are (Paragraphs 19-22, Financial Accounting Standards Board, 2001e): first, identifying potential impairments by comparing the fair value of a reporting unit to its carrying amount, including goodwill. Goodwill is not considered impaired as long as the fair value of the unit is greater than its carrying value. The second step is only required if an impairment to goodwill is identified in step one; second, comparing the implied fair market value of goodwill to its carrying amount. If the carrying amount of goodwill exceeds its implied fair market value, an impairment loss is recognized. That loss is equal to the carrying amount of goodwill that is in excess of its implied fair market value, and it must be presented as a separate line item on financial statements. 164 Evidence is not fully shown due to possible incompleteness and inconsistence of data collection, as S&P 500 companies’ disclosures on impairments are somewhat chaotic and surely not clear all times. Indeed, impairment losses may also result of provisions from SFAS 121 and SFAS 144, and many times is not clear whether the recognition of a specific impairment loss is exclusively the result of SFAS 142 adoption. In face of such situation, a conservative view has been adopted, i.e., an impairment value would only be collected if it could be associated to SFAS 142 with a high degree of confidence, and also if it could be reliably measured, in case the impairment value was not clearly identified and disclosed by the reporting company. Due to these constrains, related to the unfeasibility of verification and checking data directly with the S&P500 companies, one cannot ensure completeness of data collected, which therefore can lead to a certain degree of misinterpretation. Therefore, in this thesis the analysis made concerning impairments under SFAS 142 is a brief one.

196

S&P500 companies, not only did more companies recognise losses in 2002, as the amounts impaired in 2003 decreased significantly - they would represent only about 6% of the 2002 losses.165 Indeed, according to this thesis, S&P500 companies sample, the amount of gross impairment charges related to SFAS 142 is estimated to be as much as $169.4 thousand millions in the first year of adoption, an astounding figure, that may have exceeded the reported positive effect of nonamortisation of purchased goodwill and other intangible assets, as to be shown in the next section. This finding is consistent with seminal literature on goodwill impairment which found a “big bath earnings management” in the first year of adoption of SFAS 142 (Jordan & Clark, 2004, 2005).

Nevertheless, qualitative evidence collected in the present research suggests that impairment losses may have continued following the “big bath” in the first year of SFAS 142 adoption, in which some massive impairment losses were recorded, as shown by the cases of AOL Time Warner, $98.2 thousand millions, or Clear Channel Communications, with $11.4 thousand millions. For example, in early 2002, Adobe Inc. disclosed that (2002: 36):

“We are currently evaluating the impact of this Statement on our financial position and are planning to adopt this standard beginning in fiscal year 2003, as required. It is possible that in the future, we may incur less frequent, but larger, impairment charges related to the goodwill already recorded, as well as goodwill arising out of potential future acquisitions.”.

Afterwards, Jordan et al. (2007) would admit that the “big bath earnings management” continued throughout 2003 and 2004, a finding consistent with some indications provided by the qualitative evidence collected in the present research from S&P 500 companies’ annual filings, as companies’ executives suggested the possibility of recognition of further impairment losses.

165 Ibidem.

197

Taking once again 3M Company as an example of accounting treatment, the impairment tests procedure under SFAS 142 was described in the following way (Note 1, 3M Company, 2003):

“Beginning January 1, 2002, goodwill will be tested for impairment annually, and will be tested for impairment between annual tests if an event occurs or circumstances change that would indicate the carrying amount may be impaired. Impairment testing for goodwill is done at a reporting unit level. Reporting units are one level below the business segment level, but can be combined when reporting units within the same segment have similar economic characteristics. 3M, at year-end 2002, had 20 reporting units under the criteria set forth by SFAS No. 142. The vast majority of goodwill relates to and is assigned directly to a specific reporting unit. An impairment loss would generally be recognized when the carrying amount of the reporting unit's net assets exceeds the estimated fair value of the reporting unit. The estimated fair value of a reporting unit is determined using earnings for the reporting unit multiplied by a price/earnings ratio for comparable industry groups, or by using a discounted cash flow analysis.”.

Although the accounting procedure for impairment testing seemed to be reasonably well described, and despite the multiple business combinations involving 3M Company in the preceding years, including 9 deals in 2002, no impairment charges were recorded for purchased goodwill in 2002 (Note 1, 3M Company, 2003):

“The company completed its assessment of any potential impairment upon adoption of this standard and upon its annual assessment and determined that no impairments existed.”.

The absence of impairment charges was not an exclusive for purchased goodwill, as it was also extended for other intangible assets with indefinite life, such as patents, tradenames, and others, acquired from an independent party.166 In none

166 Below is shown the treatment given by 3M Company to purchased intangible assets with an indefinite life (Note 1, 3M Company, 2003):

198 of the fiscal years examined, from 2001 to 2004, had 3M Company recorded any impairment loss related to SFAS 142’s adoption.

This type of qualitative evidence suggests the “big bath earnings management” could have been even worse in 2002. Perhaps some companies postponed the recognition of losses due to the absence of in-depth guidance regarding the application of impairment tests. It was also an upcoming intense period of accounting regulation changes.167 Second, conceivably some companies did not have much time to carefully proceed with impairment tests, in order to evaluate any possible impairment losses under SFAS 142. In the case of 3M Company, it

“Effective January 1, 2002, with the adoption of SFAS No. 142, intangible assets with an indefinite life, namely certain tradenames, are not amortized. Intangible assets with a definite life are amortized on a straight-line basis with estimated useful lives ranging from 2 to 17 years. Indefinite-lived intangible assets will be tested for impairment annually, and will be tested for impairment between annual tests if an event occurs or circumstances change that would indicate that the carrying amount may be impaired. Intangible assets with a definite life are tested for impairment whenever events or circumstances indicate that a carrying amount of an asset (asset group) may not be recoverable. An impairment loss would be recognized when the carrying amount of an asset exceeds the estimated undiscounted cash flows used in determining the fair value of the asset. The amount of the impairment loss to be recorded is calculated by the excess of the assets carrying value over its fair value. Fair value is generally determined using a discounted cash flow analysis. Costs related to internally developed intangible assets are expensed as incurred.”.

167 The accounting changes in early 2000’s were not exclusively from business combinations, as other FASB pronouncements became effective in the same period, such as: SFAS 140, Accounting for Transfers and Servicing of Financial Assets and Extinguishments of Liabilities a replacement of FASB Statement 125; SFAS 143, Accounting for Asset Retirement Obligations; SFAS 144, Accounting for the Impairment or Disposal of Long-Lived Assets, or SFAS 145, Rescission of FASB Statements No. 4, 44, and 64, Amendment of FASB Statement No. 13, and Technical Corrections. More accounting standards would be enforced in the upcoming years. Following several scandals that led to major corporate failures, the US Congress passed the Sarbanes-Oxley Act (SOX) of 2002, enacted on 30 July, in order to improve investors’ protection from the possibility of fraudulent accounting activities by corporations.

199 was implicitly admitted that no extensive impairment tests were performed in a first stage of SFAS 142’s adoption (Notes, 3M Company, 2002):

“A preliminary review indicated that no impairment existed at December 31, 2001”.

In recent years the literature on goodwill impairment under SFAS 142 has grown substantially. Some literature casted doubts about the superiority of impairment tests over amortisations (vid. Hayn & Hughes, 2006), while other literature suggested that the managerial discretion in applying the goodwill impairment tests reduces the quality of reported earnings (e.g. Massoud and Raiborn, 2003). Nevertheless, the FASB did not change substantially the purchase method format on the years that followed SFAS 142 adoption, and therefore purchased goodwill amortisations seem to have been definitely replaced by impairment tests.

In conclusion, it is seems arguable that the replacement of amortisation of acquired goodwill, and other intangible assets with definite life, by impairment tests may have eased the recognition of impairment losses immediately upon initial adoption of SFAS 142. Indeed, not only the negative economic environment may have propelled the recognition of impairment losses, as also the positive impact from nonamortisation of goodwill and other intangible assets may have also helped impairing companies to dilute the impairments negative impact on corporate earnings. adoption and disclosure timing of SFAS 142 impacts: SFAS 142 is the most important of the new business combinations accounting standards for the present research, as the information collected from the annual reports is disclosed under its provisions. It is therefore critical to understand the process of disclosure and adoption of SFAS 142 by the companies in the USA. A brief chronology follows. After FASB issued the new business combinations accounting standards in June 2001, companies become likely to mention that fact immediately after in their annual reports. However, no impacts had to be

200 measured and disclosed immediately, as companies only needed to compulsorily adopt SFAS 142 provisions from fiscal year 2002 onwards.

Indeed, as SFAS 142 provisions were required to be applied starting with fiscal years beginning after 15 December 2001 (Paragraph 48, Financial Accounting Standards Board, 2001a), and as fiscal year ending in 31 December is the commonest financial reporting period, most S&P 500 companies adopted SFAS 142 immediately in 2002. However, companies with a fiscal year beginning after 31 December 2001 could defer adoption of SFAS 142 for 2003’s fiscal year.168

Another justification for the immediate adoption of SFAS 142 in 2002 by majority of companies, was that early application was permitted for entities with fiscal years beginning after March 15, 2001, provided that the first interim financial statements would have not been previously issued (Paragraph 48, Financial Accounting Standards Board, 2001a). Finally, it is important to recall that regardless the date of adoption, goodwill and certain intangible assets with an indefinite life acquired after 30 June 2001 would not be amortised, but tested for impairment. This means that in some cases SFAS 142 could have to be implemented in mid-fiscal year. An interesting application example is provided in SFAS 142 (paragraph 50, Financial Accounting Standards Board, 2001e):

“an entity with a December 31, 2001 fiscal year-end would be required to initially apply the provisions of this Statement on January 1, 2002; if that entity completed a business combination on October 15, 2001, that gave rise to goodwill, it would not amortize the goodwill acquired in that business combination even though it would continue to amortize until January 1, 2002, goodwill that arose from any business combination completed before July 1, 2001. Intangible assets other than goodwill acquired in a business combination or other transaction for which the date

168 It is important to note that when a fiscal year does not coincide with the calendar year, the calendar year in which the fiscal year ends is used in the shorthand. For example, if a company’s fiscal year begins in 1 February 2001, and therefore ends in 31 January 2002, it would be then considered as 2002’s fiscal year. This is the case of Wal-Mart Stores Inc., as shown a few paragraphs below.

201

of acquisition is after June 30, 2001, shall be amortized or not amortized in accordance with paragraphs 11–14 and 16 of this Statement.”.

In terms of disclosure under SFAS 142 provisions, most companies reported impacts on results for the two fiscal years preceding SFAS 142 adoption. It is important to note that companies did not measure the impact on earnings as a result of SFAS 142 adoption. For the fiscal year of adoption, companies reported the virtual impact on previously reported results instead, by means of adjusted results. The majority of the companies adopted SFAS 142 in the fiscal year 2002 and, accordingly, reported impacts for fiscal years 2000 and 2001.

A reduced number of companies did not report SFAS 142 effects for 2000 and 2001, but disclosed impacts for the following fiscal years of 2001 and 2002 only. This was the case of Wal-Mart Stores Inc. As shown in Appendix A, Wal- -Mart’s fiscal year-end is 31 January.169 According to the Form 10-K filed by Wal-Mart for the fiscal year ended in 31 January 2002 (2002: 23):

“We will apply the new rules on accounting for goodwill and other intangible assets beginning in the first quarter of fiscal 2003. Application of the nonamortization provisions of the Statement is expected to result in an increase in net income of approximately

169 The fiscal years shown in Appendix A were collected from EDGAR Online Pro database during the period 2004-2005, and do not correspond necessarily to the fiscal years exhibited in the annual reports that were examined in this research. For example, as exhibited in Appendix A, the fiscal year for NVIDIA corp. ended on 25 January. This closing date corresponds to the year 2004. However, the fiscal years in the period of analysis were slightly different, as ending dates were 26, 27, and 28 January, for 2003, 2002, and 2001, respectively. These rolling dates are the consequence of the adoption of a fiscal year that ends always on the same day of the week. In this case, some fiscal years will have 52 weeks, while a few others will have 53. Using Cisco Systems to illustrate the adoption of this particular type of fiscal year, the company announced that commencing with fiscal year 1997, the company's fiscal year would be the 52- or 53-week period ending on the last July’s Saturday (Cisco Systems, 1998). In 1997, it was a 52- -week fiscal year which ended on 26 July 1997. The fiscal 1998 was also a 52-week fiscal year, and therefore ended in 25 July 1998. The fiscal 1999 was however a 53-week fiscal year, in order to match with an ending in the last Saturday of July which, in 1999, corresponded to 31 July.

202

$250 million for fiscal 2002. Prior to the completion of the second quarter of fiscal 2003, we will complete a transitional impairment review for goodwill and indefinite lived intangible assets as of the date of adoption. Subsequently, we will perform similar impairment reviews on an annual basis. Management does not believe that the adoption of the impairment review provisions of the statement will have a material effect on the earnings and financial position of the Company.”.

It is interesting to observe that not only Wal-Mart supposedly delayed any effective decision on impairment charges under SFAS 142 to 2003’s fiscal year, avoiding the so-called “big bath earnings management” occurred in 2002, as it has also reported the impact on results for the first time only in 2002. This was due to the fact that, despite some companies were due to adopt SFAS 142 immediately in 2002, other companies were required to adopt SFAS 142 only in 2003. Indeed, when Wal-Mart filed the Form 10-K for the year ended in January 2002, the new business combinations accounting rules were already enforced, but SFAS 142 adoption was not yet effective, as only for the fiscal year ended in January 2003 was Wal-Mart required to report SFAS 142 effects.

Apart from companies that had to adopt SFAS 142 in the fiscal years 2002 and 2003, there was also the case of companies that could be entitled to adopt the new accounting standard earlier. The earliest adoption possible of SFAS 142 was for companies with a fiscal year beginning in 15 March 2001, i.e., ending in 14 March 2002. However, as stated in SFAS 142 (paragraph 48, Financial Accounting Standards Board, 2001e):

“In all cases, the provisions of this Statement shall be initially applied at the beginning of a fiscal year. Retroactive application is not permitted.”.

Apart some exceptions, not applicable in this case, this provision means that Wal-Mart could only adopt SFAS 142 in its fiscal year beginning in 1 February 2002. It cannot be then argued that Wal-Mart deferred SFAS 142 adoption. As

203

Wal-Mart could only adopt SFAS 142 in the fiscal year 2003, is has therefore disclosed information on impacts for the fiscal years ending in 2001 and 2002.

For a better understanding of this reasoning, detailed information on SFAS 142 timings of adoption and disclosure follows. Companies with fiscal years beginning between 15 March 2001 and 14 December 2001 could adopt SFAS 142 earlier, provided that the first interim financial statements would have not been previously issued. Since SFAS 142 was to be adopted unrestrictedly only from the fiscal year beginning on 15 December 2001, early adoption can be considered as “optional”.

In terms of year of adoption, all companies had to adopt SFAS 142 in the fiscal years 2002 or 2003. Companies with fiscal years ending from 14 March 2002 to 31 December 2002, could have adopted SFAS 142 for 2002’s fiscal year. Most of these companies disclosed impacts on precedent reported results for the two preceding fiscal years, i.e., 2000 and 2001. However, some disclosed impacts only for the preceding fiscal year of 2001.

Companies with fiscal years beginning from 2 January 2002 to 14 December 2002, had to adopt SFAS 142, as any deferral would constitute a violation to SFAS 142 implementation provisions. It can be then concluded that all companies had to adopt SFAS 142 at least during fiscal year 2003, in case they had not adopted it earlier. Most of companies that adopted SFAS 142 in the fiscal year 2003 have disclosed impacts on reported results for the two preceding fiscal years, i.e., 2001 and 2002. However, some companies disclosed impacts only for the preceding fiscal year of 2002.

Finally, all companies with fiscal years ending after 13 December 2003 were due to have adopted SFAS 142, as its latest adoption was required for all companies with fiscal years beginning after 14 December 2002. If a company would adopt

204

SFAS 142 from the fiscal year ending in 14 December 2003 onwards, it would be then violating SFAS 142’s adoption provisions.170 impact measurement: Companies reporting under SFAS 142 have disclosed diverse information about the impacts from the nonamortisation of acquired goodwill and indefinitely-lived intangible assets. They have also included supplemental statements of income with information about SFAS 142 impacts on both reported and adjusted bases. The measurement of SFAS 142 impacts on previously reported results, included information about the amount of acquired goodwill and indefinite-lived intangible assets not any longer subject to amortisation, and pro-forma figures for net income (or net losses), had the new accounting standard been in effect for previous fiscal years.

One of the most important figures that companies had to disclose under SFAS 142 was the impact on EPS. As a major indicator of a company’s profitability, the EPS assumes particular importance in the USA where it is highly regarded (e.g. Chant, 1980; Larcker, 2003; Marquardt & Wiedman, 2005). In fact, despite the immense variety of indicators used in financial analysis, EPS is still considered a leading indicator for evaluating share prices. For example, Huson et al. (2001) and Core et al. (2002) demonstrated that investors take dilution of

170 To summarise, companies with fiscal years beginning between 2 January 2002 and 14 March 2002 could only adopt SFAS 142 in the fiscal year 2003, while companies with fiscal years starting between 15 December 2001 and 1 January 2002 had to adopt SFAS 142 in the fiscal year 2002. Companies with fiscal years beginning between 15 March 2001 and 14 December 2001, and between 15 March 2002 and 14 December 2002, could have adopted SFAS 142 in 2002 or 2003, respectively. It is therefore possible to make the following generalisation: during the period of adoption of SFAS 142, companies with fiscal years beginning between 2 January and 14 March had to adopt SFAS 142 in the fiscal year 2003, while companies with fiscal years starting between 15 December and 1 January had to adopt SFAS 142 in the fiscal year 2002. Companies with fiscal years beginning between 15 March and 14 December could adopt SFAS 142 in 2002, provided that the first interim financial statements would have not been previously issued; or they could adopt a fiscal year later, in 2003. This generalisation does not apply in case a company had to adopt SFAS 142 during mid-fiscal year, as required by paragraph 50 of SFAS 142.

205

EPS into account when setting stock prices, while Marquardt & Wiedman (2005) suggest that managers regard diluted EPS as an important performance measure. Overall, literature continues to examine the significant implications for equity valuation from earnings per share (vid. e.g. Gil-Alana & Peláez, 2008).

The FASB requires companies’ financial statements to report EPS for each of the major categories of the income statement: continuing operations, discontinued operations, extraordinary items, and net income (Financial Accounting Standards Board, 1997a). In order to ensure the homogeneity of the sample, only comprehensive net income figures were used.

SFAS 128, Earnings per Share, specifies the computation, presentation, and disclosure requirements for EPS in the USA (Financial Accounting Standards Board, 1997a). Under SFAS 128, two formats of EPS are required to be reported: basic and diluted. This requirement was also followed in SFAS 142, as shown in its Appendix C (paragraph C5, Financial Accounting Standards Board, 2001e).

Basic EPS is computed by dividing the income available to common stockholders, in the numerator, by the weighted-average number of common shares outstanding, in the denominator (paragraph 8, Financial Accounting Standards Board, 1997a).

The diluted EPS expands on basic EPS by including the effect of all dilutive potential outstanding common shares. Basic and diluted EPS are therefore similarly computed. However, in diluted EPS calculation, the denominator is increased in order to include the number of additional common shares that would have been outstanding if the dilutive potential common shares had been issued (paragraph 11, Financial Accounting Standards Board, 1997a).

Diluted EPS is therefore a more accurate indicator than basic EPS, since it is calculated with regard to the possibility that the holders of dilutive shares exercise their shares’ options, therefore providing a more realistic picture of a

206 company’s earnings per share performance. Unsurprisingly, literature found evidence suggesting concerns of corporate executives relating diluted EPS management (vid. e.g. Bensa et al., 2003). Additionally, while some companies did not report SFAS 142 impact on basic EPS, they all have, however, reported the effect on diluted EPS. Therefore, diluted EPS has been taken into account in this research as a proxy for measuring the impact of SFAS 142 on reported earnings. resume of data collection procedure: As SEC filings were the main data source, a note on its timing follows. In terms of 10-K and 10-K405 forms, considering they need to be filed up to 3 months after the fiscal year has ended, companies with fiscal years ending from 14 March 2002 to 30 September 2002 were due to have their filings prepared in 2002. As for companies with fiscal years ending between 31 October 2002 and 30 September 2003, the 10-K forms had to be filed until the end of 2003. Finally, companies with fiscal years ending after 31 October 2003 were due to file the 10-K forms only in 2004.

In order to ensure that all companies reporting SFAS 142 effects could be identified, the SEC filings from 2001 to 2004 were examined. Although companies did not adopt SFAS 142 in the fiscal year 2001, the filings were examined in order to analyse earlier disclosures related to the new business combinations standards. This examination was also intended to capture any possible particular or abnormal disclosures. The 2004 filings were also examined, not only to collect information about companies that had to adopt SFAS 142 until the fiscal year ending in 13 December 2003 - which had to file the Form 10-K until in the first quarter of 2004 - ; but also to ensure that companies potentially delayed in reporting on new business combinations accounting could yet be included in the sample.

This additional data verification has been revealed to be important as some early disclosures on SFAS 142 would later be revised by a few reporting companies. Whereas the reported information of a more recent Form 10-K, or

207 annual report, conflicted with similar information reported in previous years, the newest information has been the one selected for the final sample. Therefore, for most companies, the search for SFAS 142 effects started with the examination of 2004’s filings, even knowing in advance that almost all companies reported such effects in 2002 and 2003 filings.

Finally, the examined period of 2001-2004 was fertile in financial reporting restatements, as several companies filed later amendments, using the Form 10- -K/A. In some cases, a half dozen of amendments were filed by a single company during the period from 2001 to 2004. It cannot be assured that every amendment filed has been fully verified, despite the confidence that the most relevant documents were checked, as the more substantive amending fillings were examined.

Additionally, despite similarities in financial reporting, some companies reported SFAS 142 effects in slightly different ways, forcing in a few occasions to rely on personal judgement in order to harmonise the information made available in annual reports by every company. It is therefore acknowledged that some data may have been mishandled, despite being examined as carefully as possible. Any inaccuracy from the data collecting process that may be reflected in the annual reports sample is exclusive responsibility of the author of this thesis.

5.3.4 SFAS 142 impacts sample

As discussed before, the sample for analysis is the result of the congregation of data collected from financial reports of companies that completed M&A deals in recent years, and that have reported business combinations accounting changes following the adoption of the new FASB’s standards. The annual filings from 2001 to 2004 of the 500 companies that composed the S&P 500 index in 2004 were examined, as this index may be considered to be a suitable proxy for a corporate population, more precisely listed companies, where a significant

208 portion of the M&A activity is placed. The use of S&P 500 index also ensures a good coverage for the most significant industries in the USA.

Telecom. Energy Services Materials 5.6% Consumer 2.0% 6.4% Discretionary 17.4% Utilities 6.6%

Consumer Financials Staples 16.4% 7.2%

Health Care Information 11.0% Industrials Technology 11.4% 16.0%

Fig. 5.5 S&P 500 index companies by industry as of 31 December 2004

As shown in Fig. 5.5, the corporate sectors included in S&P 500 index are quite diversified, being a seemingly well-weighted population. By the end of 2004, the financial and IT industries represented about one-third of S&P 500 companies. Together with consumer discretionary companies, IT and financials accounted for half of the companies in the index.

Following the examination of the 500 S&P’s companies, it has been found that more than half of the companies disclosed impacts in the scope of the adoption of SFAS 142. More precisely, at least 257 companies measured and disclosed impacts on earnings as if SFAS 142 would have been previously adopted. In addition, another 219 companies referred adoption of SFAS 142, but did not provide any measurement or details of any impact. Some other 4 companies referred the effectiveness of SFAS 142, but did not clarify whether they were

209 entitled for adoption. In resume, only 24 companies did not assume clearly SFAS 142 adoption.171

The majority of the companies disclosed impacts for 2000 and 2001, regardless of different fiscal years’ endings. However, as mentioned before, a company with a fiscal year ending between 14 March and 13 December could have adopted SFAS 142 only in 2003. In case of adoption in the fiscal year 2003, it was then natural for a company to report impacts for both 2001 and 2002.172 As a result of this myriad of possibilities, the sample comprised 257 companies with the following reporting status:

i) 224 companies disclosed information for both fiscal years 2000 and 2001; ii) 8 companies disclosed information only for fiscal year 2001; iii) 24 companies disclosed information for both 2001 and 2002; and iv) 1 company disclosed information only for fiscal year 2002.173

171 In a more accurate computation, the final sample could be reduced in five observations. Fannie Mae and Freddie Mac could have been excluded due to its “governmental” nature, together with Hospira Inc. and Freescale Semiconductor Inc. which are companies spun off in 2004 from Abbot Labs and Motorola Inc., respectively. Finally, News Corporation was registered in Australia until 2004, and did not file 10-K reports during the period of study. Nevertheless, such elimination would not have any material impact on the figures shown. 172 However, a few companies disclosing adoption of SFAS 142 in 2003 reported impacts for 2000 and 2001. 173 A small group of retailers, including Kohl’s Corporation, The May Department Stores Company, and The Kroger Co., adopted SFAS 142 in the fiscal year starting on 3 February 2002. According to them, this adoption was made in the fiscal year 2002, and therefore the impacts from SFAS 142 were disclosed for 2000 and 2001. According to the theoretical framework presented in this research, a 2003 fiscal year-end reporting corresponds to the fiscal year 2003, not 2002. This framework matches with the views of the remaining companies in the sample. Therefore, had these retailers followed the standard views on fiscal year definition, and they would have disclosed impacts for the years 2001 and 2002, as if they had adopted SFAS 142 in the fiscal year 2003. As this issue is irrelevant for the constitution of the final sample used in this research, it was decided to keep these companies in the group of companies reporting impacts for 2000 and 2001.

210

It was justified previously why companies reported SFAS 142 effects in different years. Regardless of the year of reporting, what matters for the present research is the homogenised impact on the two fiscal years preceding SFAS 142 adoption. As the large majority of the companies examined, 224 in 257, reported income figures for 2000 and 2001 together with adjusted pro-forma information as if the accounting change had been already in effect, these two fiscal years were taken as a reference for the final sample. This assumption means that data in the final sample referred to 2001 is a proxy for the impacts on the fiscal year preceding SFAS 142 adoption, i.e. -1 year, regardless the effective year of adoption by the companies. Similarly, sample data for 2000 is a proxy for the impacts on the second fiscal year preceding SFAS 142 adoption, i.e. -2 years.

In order to homogenise the reporting periods in the sample, the data from the company presenting information only to 2002 (iv) was considered as being referred to 2001; while data for the 24 companies that reported impacts for the fiscal years 2001 and 2002 (iii), has been considered as for years 2000 and 2001, respectively. As a result of this standardisation, it is implicit that all companies adopted SFAS 142 in the fiscal year 2002, despite this not being true for 25 of the 257 companies included in the final sample.

The final sample is therefore composed by 257 companies, contributing with 505 observations, 248 for 2000, -2 years, and 257 for 2001, -1 year from adoption. The weight by industries for the 257 companies included in the sample is shown in Fig. 5.6 below.

211

Telecom. Energy 2.7% Services Information Utilities 3.1% Technology 5.8% 18.7% Materials 7.0%

Consumer Consumer Staples Discretionary 9.3% 15.2%

Health Care Financials 11.7% Industrials 14.4% 12.1%

Fig. 5.6 Annual report sample companies by industry

Interestingly, the majority of companies impacted by SFAS 142 were from the IT industry, with 18.7% of the total sample. Financials ranked third, and together with IT and consumer discretionary they sum for around half of the companies in the sample.

5.3.5 Basic descriptive statistics and analysis

In terms of global figures for the sample companies, in 2000, the 248 companies reported an average net income of 890.8 millions in dollars value. On average, had SFAS 142 been made effective in 2000, the same companies would report 962.1 millions of net income, a 7.9% increase. In terms of total values for 2000, the 248 companies reported $220.9 thousand millions of net income and would report an adjusted $238.6 thousand millions had SFAS 142 been adopted, as net amortisation of purchased goodwill totalling $17.2 thousand millions, and amortisation of other purchased intangible assets than goodwill of $471 millions,

212 would be discontinued, and consequently added back to reported income. Had SFAS 142 been adopted in the fiscal year 2000, and the net income of the 246 companies would be therefore increased in $17.6 thousand millions.

In a similar analysis for 2001, the 257 companies reported an average net income of $407 millions, a 54.3% decrease when compared to 2000. As discussed earlier in this thesis, the economic climate changed in the beginning of the 2000’s, so this sharp decrease in earnings can be considered as normal. Had SFAS 142 been made effective in 2001, and the 257 companies would report $541.8 millions of net income on average, a 33.1% increase. In terms of global figures for 2001, the 257 companies reported $104.6 thousand millions of net income, and disclosed a pro-forma net income of $139.2 thousand millions, as net acquired goodwill amortisation of $32.9 thousand millions, and amortisation of $1.65 thousand millions related to other purchased intangible assets than goodwill, such as indefinite-lived tradenames, or workforce intangibles, would be added back. Therefore, had SFAS 142 been adopted in the fiscal year 2001, and the net income of the 257 companies would be increased by one-third. In face of a period of rapidly falling earnings, the immediate adoption in 2001 of SFAS 142 would be certainly welcomed by companies which were amortising purchased goodwill and other intangible assets. If no impairment losses would be recorded, the adoption of SFAS 142, by means of nonamortisation of purchased goodwill and other intangible assets, would represent a bonus of $34.6 thousand millions in earnings for the 257 companies examined. impact on diluted EPS: Since companies had to disclose the virtual impact on EPS for the immediate fiscal years before SFAS 142 adoption, and being EPS a powerful and harmonised indicator, an in-depth analysis focused on diluted EPS follows.

As disclosed pro-forma by 248 companies in 2000, the adjusted diluted net income per share was on average 20.7% superior to the diluted net income as

213 reported.174 Had SFAS 142 been adopted in the fiscal year 2001, and the reported net income of the 257 companies examined would increase 29.7% on average.175

Some companies reported a zero impact on reported EPS. More precisely, 7 companies in 2000, and 3 in 2001. Had these companies been excluded from the sample, and the average impact of SFAS 142 adoption on diluted EPS would be of 21.3% and 30%, for 2000 and 2001, respectively. On the other hand, even if one would include in the calculations 219 companies that referred SFAS 142 effectiveness but did not provide measurements or details of any possible impact, and if one would consider zero impact values for these companies, the estimated variations would yet be 16% and 11% for -1 and -2 years, respectively.176 Therefore, one can conclude that the average impacts on diluted EPS are very much meaningful by all means.

The diluted EPS sample median was 7.2% for 2000, and 9.5% for 2001. The high discrepancy of the median versus average, 7.2% vs. 20.7%, and 9.5% vs. 29.7%, suggests the existence of outliers biasing the sample average. The impact

174 The change in diluted EPS is expressed as a percentage, and shows the increase in diluted EPS had SFAS 142 being adopted in a previous reporting period. In some cases, the ratio had to be computed using negative values. In order to allow proper computations of impact using negative EPS values, the following formula with absolute values in the denominator has been employed: reported EPS− adjusted EPS tt×100 adjusted EPSt

175 The justification for the difference between the % change in diluted EPS and the % change in net earnings, probably lies mostly in the fact that the use of diluted EPS necessarily results in a reduction of basic EPS, whenever the company holds contingent shares. Taking 3M company, for example, in 2001 the increase in diluted EPS from adoption of SFAS 142 would be of 3.35%; while using basic EPS, the positive variation would be of 3.58%. Additionally, EPS values are necessarily disclosed rounded, and therefore may also contribute to increase the cleavage with reported net earnings. 176 The estimations shown are merely indicative, as there is no clear information whether such 219 companies may have had any impact for any of -1 and -2 years from SFAS 142 adoption.

214 of companies reporting zero impact is not significant, as revealed by the minor differences between the average percentages for the whole sample versus sample excluded from zero values. Therefore, significant outliers are not lower quartile, but upper quartile values.

Indeed, a substantial number of companies disclosed impressive impacts on diluted EPS, had SFAS 142 been adopted. The maximum impact on diluted EPS reported in 2000 was 4.85 times, or 385%. A total of 10 companies disclosed adjusted diluted EPS with increases of 100% or more. For 5 companies, the impact on reported diluted EPS was at least 200%. The sample standard deviation was 0.47.

Had SFAS 142 been adopted earlier, and the impact on 2001’s diluted EPS would be even more expressive. The maximum impact disclosed was 7.6(6) times the reported diluted EPS, an increase of 666%. 12 companies reported impacts of 100% or more, 6 reported increases of at least 200%, and similarly, 4 reported 400%, and 3 other, 500% or more. Unsurprisingly, the dispersion of values was higher than in 2000, and therefore the standard deviation was also higher: 0.77.

A final indication about the significant weight of the outliers follows. For 2000, 49 companies reported differences between adjusted diluted EPS and reported diluted EPS superior to the average of the 248 observations: 20.7%. In 2001, the impact for 52 companies was superior to the average impact for the 257 sampled companies: 29.6%.

5.3.6 Cross-sectional analysis

Since the analysis from the survey suggested the existence of significant effects of the new business combinations accounting standards on the IT industry, and as the sample for annual reports comprises the same 10 main sectors of activity as for the questionnaires’ sample, allowing direct data triangulation, it is

215 therefore of interest to develop a cross-sectional analysis. The cross-sectional data for analysis is shown in Table 5.1 and in Appendix D.

Table 5.1 Estimated SFAS 142 impacts on diluted EPS by industry Weighted avg. goodwill Average impact on Weighted and other intang. assets diluted EPS in avg. impact added back ($ millions)† percentage in pct.‡ 2000 2001 2000 2001 2000-01

Consumer 51.4 233.3 24.51 31.15 27.87 Discretionary (38) (39)

Consumer Staples 91.3 110.1 10.09 8.55 9.32 (24) (24)

Energy 28.1 32.0 11.97 19.43 15.71 (7) (7)

Financials 69.2 89.6 18.63 16.30 17.41 (34) (37)

Health Care 39.1 50.1 30.22 17.20 23.48 (28) (30)

Industrials 100.7 108.9 28.09 34.85 31.53 (30) (31)

Information 102.4 239.4 17.80 66.64 42.48 Technology (47) (48)

Materials 45.7 60.3 31.94 19.88 25.74 (17) (18)

Telecommunication 123.8 151.5 10.43 18.20 14.32 Services (8) (8)

Utilities 25.2 51.3 6.29 12.27 9.29 (15) (15) (Number of observations in round brackets). † Weighted average from purchased goodwill, and other intangible assets than goodwill averages. The number of observations used for computing the goodwill added back average corresponds to the number of observations used for computing the impact on diluted EPS average. The number of observations used for computing the other intangible assets than goodwill average is not shown. ‡ Weighted average for 2000 and 2001 average impacts on diluted EPS.

216

Table 5.1 exhibits the weighted average of purchased goodwill and other intangible assets that would be added back for 2001 and 2000, i.e. -1 and -2 years, had SFAS 142 been adopted. It is possible to observe that companies from industries such as energy, $30.05 millions, utilities, $38.3 millions, and health care, $44.6 millions, have the lowest average amounts of goodwill and other intangible assets; while, conversely, IT, $170.9 millions, consumer discretionary, $142.4 millions, and telecommunication services, $137.6 millions, exhibit the highest amounts, which are significantly expressive for 2001.177 In fact, the amounts of goodwill and other intangible assets rose for every industry from 2000 to 2001, but this increase was particularly significant for IT and consumer discretionary, the companies from the two industries with the highest average amounts of purchased intangible assets that could be added back under SFAS 142.

In terms of diluted EPS in 2000, utilities recorded the lowest impact, 6.29%. Companies from consumer staples, telecommunication services, and energy, with impacts of a little above 10%, were also in the group of the less possibly impacted by SFAS 142. On the opposite side, materials, 31.94%, health care, 30.22%, and industrials 28.09%, topped the ranking of the industries with the highest impacts.

Overall, impacts on diluted EPS were higher for 2001. Only consumer staples, 8.55%, was below a double-digit figure. Together with utilities, 12.27%, they were the only industries with impacts below 15%. The companies from these 2 sectors were also the ones with the lowest impacts for 2000-2001: 9.32%, and 9.29%, respectively. The highest impacts in 2001 were recorded for IT, 66.64%, industrials, 34.85%, and consumer discretionary, 31.15%. The firms from these 3 industries have also recorded the highest average impacts on diluted EPS for 2000-2001, with 42.48%, 31.53%, and 27.87%, respectively.

177 Figures shown are weighted averages for 2000-2001 and are not included in Table 5.1, as they are only available in the enclosed Fig. D.1. Vid. Appendix D for cross-sectional analysis figures.

217

Fig. D.1, shown in Appendix D, compares the 2000-2001 weighted averages for impacts on diluted EPS versus amount of purchased goodwill and other intangible assets to be added back under SFAS 142. Despite one could expect some degree of positive correlation between the average amounts to be added back and the impact on diluted EPS, this is not however clearly evidenced in Fig. D.1.178 However, it seems indisputable that energy companies and utilities disclosed low values in both categories. Conversely, IT, consumer discretionary, and industrials exhibit high values of goodwill and intangible assets, concurrently with significant impacts on diluted EPS. Additionally, as shown in Fig. D.2, companies of these 3 industries highly impacted by SFAS 142, also present the highest total values of goodwill and other intangibles to be added back. An expected outcome, as IT, consumer discretionary, and industrials, account for the highest average values of intangibles, also accounting for 3 of the 4 industries more represented in the sample.

178 This lack of correlation could mean that some companies had a significant proportion of intangible assets with finite useful lives. Indeed, there is an inverse relationship between the proportion of intangible assets with a finite useful life and the impact on diluted EPS as, in opposition to goodwill and other intangible assets with an indefinite useful life, under SFAS 142 they continued to be amortised over their useful lives, but without the constraint of an arbitrary ceiling of 40 years, as required previously by APB Opinion 17 (Summary, SFAS 142, Financial Accounting Standards Board, 2001e) - vid. also Paragraph 16 of SFAS 142 (Financial Accounting Standards Board, 2001e) which defines in detail the conditions for nonamortisation of intangible assets, and paragraphs 11 to 14 which deal with the intangible assets subject to amortisation. Therefore, earnings from companies with larger proportions of intangible assets with definite useful lives, did not benefit significantly from the nonamortisation of goodwill and other intangible assets with indefinite useful lives. However, this reasoning does not apply in this case, as figures only relate to goodwill and other intangible assets that could be added back as a result of SFAS 142 adoption. Consequently, intangible assets with finite useful lives are not considered in the figures, and can not justify the discrepancies between goodwill and intangible assets (with indefinite useful life), and impacts on diluted EPS. Such discrepancies are probably due to the fact of computations of adjusted results. In 2001, some industries were seriously affected by the economic downturn, and many companies reported net losses. Had SFAS 142 been adopted in 2001 and in many cases the impact on diluted EPS would result in a reverse to net profits. This reversal leveraged the percent increases in diluted EPS figures. It also justifies the existence of additional outliers in the dataset (vid. next footnote).

218

Considering that the impacts from 2000 and 2001 could be extrapolated for 2002 onwards, seems justifiable to argue that companies from IT, consumer discretionary, and industrials, were the most affected by SFAS 142 adoption. The overall evidence also indicates that IT companies were the most impacted by SFAS 142. Indeed, not only did they exhibit the highest average values of purchased goodwill and other intangibles subject to nonamortisation in 2001, as they were also the ones suffering the most significant impacts on diluted EPS.

Finally, as referred earlier, evidence presented in this section was not subject to formal statistical testing and validation, and therefore needs to be carefully considered. An example of the possible consequences of missing control for statistical assumptions follows. The company with the highest impact on diluted EPS in the sample is from the IT industry, 666% in 2001. Despite the sizeable number of IT companies in the sample, 48 in 2001, had this outlier been removed and the average diluted EPS in 2001 would have decreased from 66.64% to 53.88%. Similarly, the average value of purchased goodwill and other intangibles would decrease from $201.6 millions to $201.4 millions. An insignificant impact in this case, however. Overall, the elimination of this outlier, or unusual observation, does not change fundamentally any of the analyses drawn before. However, since it reduces significantly the impact on diluted EPS, it therefore smoothes the prevalence of IT over the remaining industries in what concerns to SFAS 142 effects.

Other major outliers, e.g. impacts over 200% on diluted EPS in 2000 and 2001, are included in health care, financials, industrials, and IT industries.179 Although these industries are among the ones with the highest number of observations, the corresponding figures and findings need however to be regarded with due care, particularly for single yearly analyses.

179 Outliers’ threshold set arbitrarily, but certainly corresponding to the upper quartile of any sample when considered at the sub-level of different industries.

219

Weighted average figures, including both 2000 and 2001, are more robust, as the potential effect from possible outliers is more diluted, providing therefore more reliable analyses. Taking again the IT industry as an example, had the same outlier been removed and, for the period 2000-2001, the weighted average diluted EPS would be reduced from 42.48% to 35.61%, while the weighted average value for goodwill and intangibles added back would have an imperceptible decrease from 144.42 millions to 144.38 millions.

Although not shown, further sensitivity analyses were performed to ensure that the findings presented in this section were not significantly biased, and to minimise the lack of statistical testing and validation of the results obtained.

5.4 Conclusions

In this chapter, diverse evidence collected from S&P 500 companies, by means of a survey and analysis of annual reports, was examined and triangulated, whenever appropriate, in the scope of existing literature review. Following some cross-sectional analysis from both survey and annual report analysis, the overall evidence suggests that, at least, the IT industry may have suffered some significant impacts from the changes in the accounting regulation. Indeed, the concerns from some industries, with a particular relevance for IT and financials, that the new accounting regulation would damage the M&A activity, were made visible in the survey.

Whether such effects were expressive enough to impact M&A activity on IT and other industries is not definitive, as evidence can be regarded as mixed. For example, it was not possible to find a plausible justification for a very low response rate from the IT industry - the lowest from the seven industries that replied to the questionnaire. This apparent missing interest may suggest overall satisfaction with the new business combinations accounting standards. The low percentage of participation in the survey also resulted in a very small sample for

220 the IT industry, which implicitly subdues the testing made for this industry alone.

From the analysis of 10-K forms and annual reports it was possible to find that the accounting changes produced significant impacts in the financial reporting. The difference between adjusted diluted EPS and reported diluted EPS was 20.7% in 2000, and 29.6% in 2001. These effects are sizeable enough not to be neglectful of. Moreover, such impacts represent billions of dollars in purchased goodwill and other intangible assets not to be amortised that, if not subject to significant impairment losses, mean meaningful earnings increases simply as a result of a technical adjustment, i.e., a change in GAAP. Additionally, such positive impacts were particularly relevant for the IT industry, with a 66.64% increase in diluted EPS in the -1 year of SFAS 142 adoption, and an average 42.48% increase for -1 and -2 years, the highest values among all industries composing the S&P 500 index. This reality is a normal outcome of the fact that IT companies are the ones having the highest values of goodwill in their balance sheets among the entire S&P 500 sample.

Overall, it seems the new M&A accounting policy benefited companies, with a particular emphasis for the ones from the IT industry, and also for other industries, such as or consumer discretionary, or industrials. However, on the other side of the equation, one must not forget that the positive effect of nonamortisation of purchased goodwill and other intangible assets was likely largely outdone by the “big bath” earnings occurred immediately in the first year of SFAS 142 adoption. In the case of the IT industry, the impairment charges in the sample were estimated to be as much as $7.3 thousand millions. Not the highest value among S&P 500 industries, as consumer discretionary estimated value of gross impairment charges are estimated to be some impressive $124.1 thousand millions, with the notable contributes of large losses from Clear Channel Communications, Nordstrom, and Time Warner. It therefore appears that IT firms restrained the recognition of impairment losses during the “big bath” earnings period. Indeed, the $7.3 thousand millions recognised in the first year of adoption, would be largely compensated by the

221

$11.5 thousand millions resulting from nonamortisation of purchased goodwill and other intangible assets whether SFAS 142 would have been adopted in -1 year.

The impacts of the accounting regulation should not be attended exclusively for the IT industry. For example, consumer discretionary companies recognised massive amounts of impairment losses. But there are other industries largely affected, as the case of industrials, which have similar values of goodwill when compared to the IT industry, and also very significant positive effects on earnings resulting from the nonamortisation of goodwill and other assets.

In conclusion, from 1st July 2001 onwards, firms can no longer avoid the goodwill burden, and events such as the “big bath” earnings of 2002 warned companies about the perils of being involved in M&A deals, particularly when involving large values of purchased goodwill. These worries may also support allegations that the M&A accounting changes could negatively impact the financial reporting of companies accounting for business combinations, possibly inhibiting M&A announcements and the completion of deals, therefore harming the USA economy dynamics. The discussion of the findings triangulation started in this chapter is to be continued in the later chapters of this thesis. Together with evidence brought from previous chapters, the research findings are to be related to existing literature, whenever appropriate.

222

Chapter 6 Data Collection

6.1 Introduction

In the context of the literature review made in this thesis, together with the subsequent examination of diverse evidence about M&A activity and business combinations accounting, and following the development of the research hypotheses, accompanied by the explication and justification of the methodology for analysis, this chapter shows the data collection procedure, together with the selection of the final samples to be used in the research models, which will be constructed and subject to testing in the next chapter.

There are many M&A data providers and therefore this chapter starts by analysing some different data sources. Based on this analysis, the justification for the selection of a particular M&A database is presented.

Following the selection of the data source of M&A to be used in this thesis, this chapter describes the data selection criteria, which was directed by four main vectors: deal timing, type of deals, deal status, and measure of M&A activity. The criteria adopted for each vector are supported by means of technical definitions from the M&A database, and by the due appropriateness for this research purpose.

Finally, following the presentation of the final samples to be used in the research models, a wide variety of different types of data aggregation used in

223

M&A literature is shown. The constitution of aggregated sets of data can improve the testing of the research hypotheses, as justified by means of literature review.

6.2 Data sources

Selecting an appropriate dataset is critical when developing research in the M&A field. This is particularly true for studies using long time series, since the inexistence of a continuous, and consistently assembled, time series on the number of M&A deals result in difficulties to researchers (Town, 1992).180 This need for assembling in order to obtain larger sets of data involves a trade-off with the consistency of the time series (Golbe & White, 1993). In general, the problem is missing information, although sometimes the difficulty is to choose between different sources of data and estimations.181

Nowadays, the scenario is different, because there are several reputable companies which track, produce, and publish reliable statistics for M&A, such as Thomson Financial, Mergerstat, Bloomberg, or big accounting and consulting firms, such as KPMG in association with Dealogic. Therefore, assembling and estimating data is no longer an issue. However, selecting a suitable data provider remains a key concern.

Among the different tracking companies, there are two which have produced comprehensive statistics for M&A activity in the USA for many years: Mergerstat and Thomson Financial (henceforth referred as Thomson).182 Their

180 Until recently, the most relevant M&A datasets were provided by Nelson, 1895-1920, Thorp, 1919-1939, FTC (U.S. Federal Trade Commission), 1940-1979, and the periodical Mergers & Acquisitions, since 1967 (Golbe & White, 1993; Town, 1992; Weston et al., 1990). 181 For example, the number of industrial consolidations during the first merger wave, 1897- -1904, was differently estimated by Moody, Watkins, Thorelli, and Nelson (Banerjee & Eckard, 1998). Golbe and White (1988), present a summary of time series merger data and their limitations. 182 More recently, Thomson Financial merged with Reuters creating Thomson Reuters.

224 data is frequently used in research papers and books. Mergerstat produces statistics since 1963.183 It covers transfers of ownership of at least ten percent of a company’s equity, priced at a minimum of US$ 1 million, and involving at least one entity from the USA (Mergerstat, 2003). The other main data provider is Thomson, a company which has expanded its activity in recent years and possesses a wide range of publications and databases directly concerned with M&A activity, such as Mergers & Acquisitions, Acquisitions Monthly, Thomson Deals, and SDC Platinum. Thomson owns data since 1967.184 Nevertheless, Thomson’s main database, SDC Platinum, a former Security Data Company’s online database of M&A, only covers deals since 1977 (Thomson Financial, 2006).185 Both Mergerstat and Thomson collect data for publicly traded, privately owned, and foreign (non-US) companies.

Table 6.1 Summary of some major M&A data sources for the USA

Mergers & SDC Platinum Mergerstat Acquisitions

Minimum value $5 million (lower $1 million (lower minimums in the No minimum minimums in the for tracking past) past)

Geographical U.S.A. U.S.A. & U.S.A. range Global

Period Since 1967 Since 1977 Since 1963

M&A and M&A and Type of deals Divestitures Miscellaneous Divestitures

Data sources: Mergerstat (2003), Thomson Financial (2002; 2006).

183 Early publications by W.T. Grimm. 184 Early publications by Mergers & Acquisitions. 185 Until the mid-1980’s the database only covers a few dozens of transactions which cannot be considered even as a mere sample of the M&A activity. From early 1980’s, the coverage appears to be comprehensive, nevertheless, according Pryor (2001), it should be regarded more as a sample. According to Thomson Financial (2006), only after 1990 can the completeness of the data be ensured (Pryor, 2001).

225

The tracking quality varies from company to company, as also the criteria for classifying the operations, and the methodology for valuating the transactions.186 Despite some inaccuracies, Mergerstat and Thomson both seem to provide a good coverage and a good quality of information for the period after 1990. Fig. 6.1 shows the evolution of the number and value of net announced M&A deals, according to Mergerstat, and the number and value of completed M&A deals, according to Thomson’s SDC Platinum, since 1990.187 Divestitures, such as sales of corporate units, unit managements buyouts, and minority equity interest purchases (Mergerstat, 2003), are not included.

$ in Thousand Millions No of Deals

1600 10000 Value (SDC) 1200 7500 Value (Mergerstat) 800 5000 No. Deals (SDC) 400 2500 No. Deals (Mergerstat)

0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Fig. 6.1 Yearly number and value of M&A deals in the USA, 1990-2002 Data sources: Mergerstat, for Net Announced deals, and SDC Platinum, for Completed deals (Mergerstat, 2001, 2002, 2003; Thomson Financial, 2006).

Undoubtedly, the M&A activity peaked in between 1998 and 2000, the whole triennial being very intense. However, looking more carefully at Fig. 6.1, it is possible to observe that, in terms of the number of deals, the M&A activity peaked differently depending on the database taken into account: roughly in

186 e.g. whenever payment includes shares, debt, combination, or other than cash, valuation may vary. 187 “Net announcements” stand for total announcements less cancellations and competing bids (Mergerstat, 2003). Unlike Thomson, Mergerstat does not use the concept of “completed deals”.

226 between 1998 (Thomson Financial, 2006), and 1999-2000 (Mergerstat, 2003). This discrepancy of data is also registered when the peak activity is measured by dollar value: in between around 1999 (Mergerstat, 2003), and 2000 (Thomson Financial, 2006), approximately.

The comparison of Mergerstat’s and Thomson’s M&A series also allows to conclude that SDC platinum coverage it is significantly superior to Mergerstat. This was expected, since Mergerstat only tracks M&A deals above US$ 1 million. Furthermore, from the literature review made, it is the author’s perception that the majority of contemporaneous studies utilise SDC Platinum data (e.g. Harford, 2005; Rhodes-Kropf et al., 2005). Hence, the M&A data from Thomson’s SDC Platinum has been selected to the development of the models used in this thesis, as shown in chapter 7.

Additionally, macroeconomic and financial market data for the USA have been collected from different sources. Stock prices indexes, GDP, and interest rates data have been selected from Compustat database, and Datastream; market capitalisation from World Federation of Exchanges; industrial production from the Federal Reserve Board of the USA; and information about financial market holidays, and federal holidays, from NYSE, and the US Office of Personnel Management, respectively.

6.3 M&A data selection

As will be described throughout the remainder of this chapter, there are two main samples to be used in the models development, in chapter 7: one comprising monthly, weekly, and daily data, for the 2000-2002’s period; and another one using quarterly data, for the 1994-2008’s period.

M&A transactions generally involve complex contours, as a consequence of a myriad of choices and limitations at different levels, namely: financial, strategic, managerial, legal, regulatory, accounting, and fiscal. Diverse information is

227 available at SDC Platinum about US target enterprises and respective deals, such as: date of announcement and date of completion, status of deal, percentage of shares acquired, value of deal, and target firm value. Thus, it is necessary to sort the relevant M&A data, in order to be possible to test the hypotheses previously formulated. Accordingly, the data selection criteria involve four main vectors, which can be identified as follows: deal timing: There are two major moments during the development of an M&A deal. The first one is the announcement date, which is when either acquirer or target firms, alone or together, publicly announce that they hold negotiations, or that they have received a formal M&A proposal. The other is the effectiveness date, which is when the transaction is formally declared completed, or withdrawn.

The date of announcement is a critical issue in the present study, as discussed in chapter 4. In the case of the sample that served to aggregate monthly, weekly, and daily M&A data, the inclusion criteria imply that all deals must have been announced between 2000 and 2002, regardless of the date of effectiveness. This means that deals announced before 2000 are not included at any circumstance, even if they have achieved later completion between 2000 and 2002. Conversely, deals announced during that triennial period, and made effective, i.e. completed, only after the end of 2002, are included. The justification for this procedure relies on the fact that the timing of M&A occurrence corresponds to its announcement, while its effectiveness only matters afterwards, in order to confirm the deal status: completion or cancellation. The same rationale serves for the 1994-2008’s sample of quarterly data, by simply replacing 2000 by 1994, and 2002 by 2008.

In order to simplify the terminology used, from this point of the thesis only the expression “announced deals” is to be used, instead of “announced and completed deals”. This simplification results from the adoption of the view that any announced deal to be included in the sample must have been effectively completed until 2006, in the case of the 2000-2002’s sample; or until 2009, in the

228 case of the 1994-2008’s sample, which is the last time when M&A data in SDC Platinum has been verified. type of deals: Being a very comprehensive database, SDC Platinum covers a wide range of operations related with corporate restructurings, such as: Spin-Offs and Equity Carve-Outs, Exchange Offers, Repurchases, Recapitalisations, Self-Tenders, Minority Purchases, Acquisitions of Remaining Interest and Privatisations.188 As this thesis regards M&A in the strict context of takeover and control exchange, M&A deals tracked by Thomson have been selected only whenever the acquiring firm was purchasing an interest of 50% or over in a target firm, raising its interest from below 50% to above 50%. Successful tender offers and

188 Deal types definitions (adapted from Thomson Financial, 2006): Spin-Offs: a company tax free distribution of shares of a unit, subsidiary, division, or another company’s equity, or any portion thereof, to its shareholders; Recapitalisations: a deal which is a recapitalisation, or a deal which is part of a recapitalisation plan, where the company issues a special one-time dividend in the form of cash, debt securities, preference shares, or assets, while allowing shareholders to retain an equity interest in the company; Self-Tenders: comprises all deals in which a company announces a self-tender offer, recapitalisation, or exchange offer. In a self-tender offer, a company offers to buy back its equity securities or securities convertible into equity through a tender offer. A company essentially launches a tender offer on itself to buy back shares; Exchange Offers: a deal where a public company offers to exchange new securities for its outstanding securities; Repurchases: includes all deals in which a company buys back its shares in the open market, or in privately negotiated transactions, or a deal where a company’s board authorises the repurchase of a portion of its shares; Minority Purchases: contains all deals in which a company is acquiring a minority stake (i.e. up to 49.99% or from 50.1% to 99.9%) in the target company; Acquisitions of Remaining Interest: embraces all deals in which a company is acquiring the remaining minority stake (i.e. from at least 50.1% ownership to 100% ownership), which it did not already own, in a target company. The acquiring company must have already owned at least 50.1% of the target company and would own 100% of the target company at completion; Privatisations: a government or government-controlled entity sells shares or assets to a non- -government entity.

229 leveraged buyouts (LBO) have been also included. The justification for this inclusion follows in the next two paragraphs. All other types of deals, as described in footnote number 188, were therefore excluded from the analysis.

Unlike mergers, which are in general friendly agreements between the management of the bidding and target firms, tender offers involve the purchase of shares without the need for approval from the target’s Board of Directors. Some tender offers may have the approval of the target management, while, conversely, others may be regarded as being hostile. Thomson understands a tender offer as a formal offer of definite duration made to the equity holders, in order to purchase shares from public company. The offer is often conditioned upon certain requirements, such as a minimum number of shares to be tendered. According to the US Securities and Exchange Commission (2003), a tender offer is a broad solicitation, made for a limited period by a company or a third party, to purchase a substantial percentage of a company’s shares or units. The offer is at a fixed price, usually at a premium over the current market price, and is contingent on shareholders tendering a fixed number of their shares or units (Securities and Exchange Commission, 2003). Tender offer rules generally do not apply to bids that result in the ownership of less than five percent.189 In terms of tender offer characterisation, both Thomson’s and SEC’s glossaries seem to coincide, which means that SDC Platinum only considers offers resulting in a minimum of five percent in ownership. Because these operations usually involve the purchase of a substantial percentage, generally leading to changes in ownership, tender offers at SDC Platinum database were included.

A LBO occurs when an investor group, investor, or a generic firm, offers to acquire a company, taking on an extraordinarily large amount of debt, with plans to repay it with funds generated from the target company, or with revenue earned by selling off assets from the newly purchased company. Thomson considers a deal to be an LBO, if the investor group includes

189 Also known as “mini-tender offers”. Bidders in mini-tender offers usually limit the offer to below five percent, so that they do not have to comply with many of the investor protections that are in place for larger tender offers (Securities and Exchange Commission, 2000).

230 management of the target firm, or in case the transaction is identified as such in the financial press and one hundred percent of the company is acquired. Since this type of deals result in a change in the effective control of the companies, they were also included. deal status: In terms of deal status, tracking companies, such as Thomson and Mergerstat, can capture the picture of the deal at very different stages, from a mere rumour or intention, up to its completion.190 The majority of the studies focused on M&A use completed deals (e.g. Conn et al., 2005; Melicher et al., 1983; S. B. Moeller et al., 2005), while a few others employ or refer to announced deals (e.g. Black, 2000a, 2000b; Rhodes-Kropf et al., 2005). Since the main purpose of the present research is to investigate any effects of the FASB changes to the factual M&A activity, the analysis is focused only on announced deals that were effectively completed or withdrawn afterwards. Consequently, the following SDC Platinum’s deal status were not selected: rumoured, discontinued rumour, intended, intended acquisition withdrawn, partially completed, seeking buyer,

190 Deal Status definitions (adapted from Thomson Financial, 2006): Completed: the transaction has closed; Withdrawn: the target or acquirer in the transaction has terminated its agreement, letter of intent, or plans for the acquisition or merger; Pending: the transaction has been announced but has not been completed or withdrawn; Partially Completed: the tender offer has been completed, but the actual merger of the two parties has not yet taken place; Intended: the acquirer has announced that they propose or expect to make an acquisition (generally used for repurchases); Seeking Buyer: the target company has announced plans to seek out a buyer or buyers for its assets or the company itself; Seeking Buyer Withdrawn: the target in the transaction has terminated its plans to seek out a buyer or buyers for its assets, shares, or the company itself; Rumour: reports about a likely transaction have been published in the media, but no formal announcement has been made by either the target or acquirer firms; Discontinued Rumour: a target company has formally denied the rumour of an acquisition or merger; Unknown: no definitive, conclusive evidence of the outcome of the deal was available after extensive research.

231 seeking buyer withdrawn, unknown, and pending. One potential issue is that there is a very large number of pending deals in the period from 2000 to 2002. Indeed, SDC Platinum seems to have failed to follow through about fifteen percent of all announced deals.191 This situation may configure a deal status misclassification. Despite the possibility of some, or even many, of these pending deals have been completed in the meantime, it has been decided to exclude them, keeping a strict completion criteria, as Thomson was unable to confirm that transaction effectiveness until 2006. Although the deals’ status was not comprehensively examined for this larger period of analysis, a similar finding can possibly be drawn for the 1994-2008’s sample as, until recently, SDC Platinum database continued to include a considerable number of pending deals (Thomson Financial, 2006). measure of M&A activity: Previously, in Fig. 6.1, M&A aggregated data is exhibited in the two available forms: number of transactions and deals’ value. Some literature suggests (e.g. Golbe & White, 1988; Scherer & Ross, 1990; Town, 1992), or uses (e.g. Nitzan, 2001), the dollar value of deals as the best measure for M&A activity. However, the majority of studies reviewed adopted the number of deals as a metric for M&A activity (see e.g. Barkoulas et al., 2001; Golbe & White, 1993; Linn & Zhu, 1997; Melicher et al., 1983; Town, 1992).

The use of dollar values raises several issues. It is worth remembering that M&A deal valuations may vary depending on the tracking company criteria, and also on the valuation method used. Nevertheless, Thomson’s methodology seems appropriate. According to SDC Platinum, the transaction is valuated by the total value of consideration paid by the acquirer, excluding fees and expenses. The consideration includes the amounts paid for all common shares, common shares equivalents, preference shares, debt, options, assets, warrants, and stake purchases, made within six months of the announcement date.192 Any

191 Author’s estimation (data source: Thomson Financial, 2006). 192 In the case of common shares, if a portion of the consideration paid uses this mean of payment, the shares are valued using the closing price on the last full trading day prior to the

232 liabilities assumed are also included only if they are publicly disclosed (Thomson Financial, 2006).

Many deals included in the SDC Platinum database have a disclosed dollar value.193 Wisely, Town (1992) points out that the real value of merged firms would be the ideal measure of M&A activity, but at the same time he has also identified the potential negative consequences of its use. In fact, one would agree that a M&A deal of thousands of millions dollars cannot be compared in importance with one valued in a few thousands only. However, in the period between 2000 and 2002, more than half of the completed deals have an undisclosed value.194 Moreover, the deal value is frequently estimated from publicly accessible information, which in general results in an understating of the true value of the transaction, particularly when the negotiations’ details are kept undisclosed. This understating error is made visible at SDC Platinum for the years 2000 to 2002: it is possible to compare the value of the deal and target firm value of only a few transactions, but in the majority of the cases the value of the target firm is higher than the value of the deal, an unexpected outcome, since generally M&A involve bid premiums.195 This suggests the existence of a relevant data bias.

To resume, additionally to the fact that, in most cases, M&A deals are not disclosed, it is also common that M&A deals include other than cash payments (e.g. shares, options, warrants), often resulting in large differences in terms of deal valuations among M&A tracking companies, with significant consequences in terms of accuracy of the databases. Consequently, similarly to most studies in the M&A area, the present research uses exclusively the number of completed deals as being a more reliable M&A activity proxy. announcement of the terms of the shares’ swap. For public target 100% acquisitions, is used the number of shares existent at the announcement date (Thomson Financial, 2006). In the case of preference shares, they are only included in the deal value if they are being acquired as part of a 100% acquisition (Thomson Financial, 2006). 193 Equals total value of consideration paid by the acquirer, excluding fees and expenses. 194 Author’s estimation (data source: Thomson Financial, 2006). 195 Ibidem.

233

6.4 M&A sample

Following the sample selection procedure for this thesis, a transaction recorded at Thomson’s SDC Platinum online database will be included in a sample if it satisfies the following criteria: (1) The transaction is either a merger, acquisition, LBO, or a tender offer that may lead to a change in the control of the target firm; (2) The deal was announced during the period from 1 January 1994 until 31 December 2008;196 (3) The M&A was successfully completed, or formally withdrawn.

Following these sample selection criteria, in order to construct the main supporting variables for the models that were developed to test the research hypotheses, two main samples were assembled as follows. sample for the 2000-2002’s period: This sample comprises announced deals involving US target firms during the period between 2000 and 2002. According to SDC Platinum (Thomson Financial, 2006), during this period a total of 24,670 M&A deals, with a disclosed dollar value of 3.15 trillion, were announced. Table 6.2 summarises the sample construction for the 2000-2002’s period.

196 When referring to the sample for the 2000-2002’s period, the timing of announcement is, obviously, from 1 January 2000 until 31 December 2002.

234

Table 6.2 Sample description for the 2000-2002’s period Value Deal Status Number of Deals ($ millions)a 2000-2002 announced M&A in the U.S.A. 24,670 3,151,924.6 Less Intended (77) (2,573.2) Intended Acquisition Withdrawn (7) (1,391.0) Partially Completed (7) (335.5) Discontinued Rumour (17) (0.0) Rumoured (54) (0.0) Seeking Buyer Withdrawn (23) (0.0) Seeking Buyer (240) (0.0) Pending (3,581) (4,006) (34,981.2) Total sample size 20,664 3,112,643.7

Completed deals 19,758 2,734,269.7 Withdrawn deals 906 378,374.0 Final Sample 20,664 3,112,643.7 aTotal deal value, including Net Debt of Target. Total amounts only for deals where value information is available. Data source: SDC Platinum (Thomson Financial, 2006).

The selection process resulted in the elimination of 4,006 deals, which were either pending, or unconfirmed (intended, rumoured, etc). The value of these exclusions is significantly less important, as it totals about 35 thousand millions of dollars. The final sample consists of 19,758 completed transactions and 906 withdrawn deals, with total dollar values of 2.7 trillion and 0.378 trillion, respectively. sample for the 1994-2008’s period: Similarly to the process of construction of the 2000-2002’s period sample, a larger set of M&A data was collected in order to assemble a sample for the 1994-2008’s period. This larger sample is composed by a total of 122,871

235 completed deals, with a disclosed value above $14 trillion; and a total of 5,017 withdrawn deals, summing almost $2.5 trillion.197

Table 6.3 Sample description for the 1994-2008’s period Value Deal Status Number of Deals ($ Millions)a

Completed deals 122,871 14,414,947.7 Withdrawn deals 5,017 2,409,106.5

Final Sample 127,888 16,824,054.2 aTotal deal value, including Net Debt of Target. Total amounts only for deals where value information is available. Data source: SDC Platinum (Thomson Financial, 2009).

6.5 Data aggregation

There is a wide variety of types of data aggregation in M&A studies. Literature focused on M&A activity tend to use annual and quarterly data, while studies on M&A performance use preferably monthly, weekly, or daily data. annual and quarterly data: A substantial number of studies based on M&A determinants, or on the M&A wave pattern, have used annual data. This is not exclusive to seminal studies (e.g. Eis, 1969; Markham, 1955; Weston, 1961), as later works have also relied on annual data (e.g. Golbe & White, 1993; Jovanovic & Rousseau, 2002; Shughart II & Tollison, 1984). The present author believes that the use of annual data is mainly justified by two reasons: limitations on early M&A datasets, and nature of the studies. As mentioned before, early M&A data is, to some extent, flawed, as it suffers from the absence of continuously and consistently assembled time series (Town, 1992). The other main reason that

197 Tallying in the UK and in Europe to $14.4 billion, and $2.4 billion, respectively.

236 also justifies the use of annual data, particularly on later studies, is the nature of the studies themselves. In some cases, the use of annual data may be more suitable, accordingly to the methodology adopted.

Nelson’s (1959) work has improved the consistency of early M&A data, and allowed the use of quarterly data. Ever since, the use of quarterly data has increased in M&A activity studies (see e.g. Barkoulas et al., 2001; Melicher et al., 1983; Town, 1992). monthly, weekly, and daily data: Early literature based on M&A performance, measured by accounting or market returns, tended to use preferably monthly (e.g. Elgers & Clark, 1980; Ellert, 1976; Hong et al., 1978; Langetieg, 1978; Leeth & Borg, 2000; Mandelker, 1974), or, less frequently, weekly returns (Banerjee & Eckard, 1998; M. L. Davis, 1990; Fabozzi et al., 1988), to estimate abnormal returns around the M&A announcement date.198 More recent studies rather tend to use daily data (e.g. Aktas et al., 2004; Ayers et al., 2002; Bradley et al., 1983; DeLong, 2003; Dodd, 1980; Duggal & Cudd, 1998; Fuller et al., 2002; Hazelkorn et al., 2004; Hubbard & Palia, 1999; Matsusaka, 1993; Mitchell et al., 2004; Moehrle et al., 2001; S. B. Moeller et al., 2005; Rosen, 2006; Weber, 2004), or use simultaneously both monthly and daily returns (see e.g. Norris & Ayres, 2000). Finally, studies on cancelled deals also employ the event study methodology, using weekly (e.g. Fabozzi et al., 1988), or daily returns (see e.g. Bradley et al., 1983; Davidson III et al., 1989; Davidson III et al., 2002; Dodd, 1980).

Several authors argue that monthly data is more consistent than data collected on a daily basis (e.g. Kennedy & Limmack, 1996). This is particularly true for long-term studies. As to the advantages of the use of monthly data over the use of daily data, Fama (1998) and Mitchell & Stafford (2000) point out that: (i)

198 This sort of literature is not only concerned with accessing returns by itself, this is, computing returns may not be an end itself, but a mean to achieve other findings. For example: the analysis of M&A returns can be used, or related, as for a proxy of M&A activity, or for examining other specific effects, such as impacts from discretional accounting policies.

237 the average monthly abnormal returns are more likely to match the model of expected returns; and (ii) the distribution of monthly returns is approximated by the normal distribution, allowing classical statistical inference, while daily returns often follow an abnormal distribution (Berry et al., 1990). In addition, Brown & Warner (1985), addressing the issue of the disadvantages of the use of daily data, highlight the (iii) problem of non-synchronous trading; and Blume & Stambaugh (1983) and Conrad & Kaul (1993) also refer to the (iv) potential overstatement of the size of excess returns.

To resume, literature on M&A determinants and on wave pattern uses the number or the value of M&A deals, both announced or completed, and mostly aggregated in annual or quarterly series; while studies based on M&A performance typically use shorter periods: monthly, weekly and, preferably, daily data, despite more possibly challenging from a methodological point of view. Daily M&A returns are easily obtainable from stock market databases, such as Datastream, Compustat, or CRSP, allowing very-short term impact analysis, following the occurrence of M&A announcements.199

This thesis’ research presents some specific characteristics and objectives, as so does its data aggregation as a consequence of such aims. By analysing the impact of the changes in accounting rules on the M&A activity, this thesis uses M&A completed and withdrawn deals as proxies for the study of M&A activity.200 Furthermore, by using two main event study’ periods, with distinct time spectrums of analysis, one relatively short, only three years, from 1st January 2000 to 31th December 2002; and also a larger one, from 1994 to 2008, this thesis has also considered diverse data aggregations. Likewise, many studies focused on M&A activity and on M&A performance, data was quarterly aggregated for 1994-2008, and aggregated on monthly, weekly, and daily sets, for the 2000-2002’s period.

199 CRSP database, or Center for Research in Securities Prices, a database administrated by the University of Chicago, Illinois, USA. 200 Being M&A returns regarded like a proxy for stock prices movements on M&A deals.

238

The use of daily data is critical to investigate whether the accounting changes resulted in any impacts in the short-term. The use of daily M&A activity also makes sense from a theoretic and methodological point of view, as to be further discussed in chapter 7. The M&A activity is historically important in the USA, and is by far the most active market worldwide, ensuring many thousands of deals per year in recent decades (vid. e.g. Black, 2000a, 2000b; Kummer & Steger, 2008; Muehlfeld et al., 2007). This ensures a regular daily stream of M&A deals in ordinary market conditions in the USA, even when facing a mild economic recession (vid. e.g. Fig. 4.1, Kummer & Steger, 2008). It is a fact that, in the USA, M&A deals are continuously made on a daily basis, just like financial markets are operating during working days, but with the advantage that a company may announce a deal on any non-working day.201

Despite the numerous warnings that can be found in literature about the importance of preparing carefully a M&A deal, including a proper ex ante planning (vid. e.g. Reed et al., 2007), the reality is that, besides made on a daily basis, sometimes M&A deals are decided in only a few days time, often neglecting the contextualisation in a long-term strategic view, increasing therefore the odds of a future failure (Kummer & Steger, 2008; S. Moeller & Brady, 2007).

It is also important to bear in mind that announcing a deal is a fair light decision, at least when compared with deciding about a final agreement with the target or merging firms, which can reveal to be complex (Kummer & Steger, 2008). Moreover, short-term strategy is used many times in M&A activity as a quick reaction, which is often critical in order to cope with other competitors and rival companies also interested in the same target or merging companies (Bradley et al., 1983).

Similarly, daily occurrences often change the markets conditions, and may affect decisions concerning M&A deals. As mentioned earlier, M&A activity is highly correlated with some market factors, such as stock prices moves (e.g. Melicher

201 Even if M&A announcements are unlikely at weekends, as discussed in Chapter 4.

239 et al., 1983). To illustrate this rationale, an example follows. Subsequently to a takeover bid, it is natural that not only the market price of bidder(s) and target(s) companies react, but that other firms from the same bidder and target industries react as well, based on the speculation that more deals may occur in the industry. Moreover, the market as a whole may be influenced, leading to a ‘merger fever’, or ‘merger mania’, that may result in a M&A wave (Gaughan, 2002). Such environment of increasing market movements often instigate rival companies to launch bids that can be decided overnight and formally executed in a few days’ time, dismissing any proper planning and valuations, simply because sometimes speed and timing are everything in M&A activity, as M&A actors and advisers are well aware. For example, Bank of America Business Capital (2006) acknowledges it while quoting Beuche (2006):

“There’s been a resurgence of domestic and international mergers and acquisitions activity. With it has come increasing pressure on business development people, investment bankers, accountants and attorneys to get deals done quickly.”.

This need for deals to be done quickly is far from new. They are the so-called “quick mergers”, i.e. quick M&A agreements that often involve mergers of companies with unrelated businesses, a strand of deals that can be dated from the first great merger movement time (Gaughan, 2007). Examples of “quick mergers” include the acquisition of Capital Cities/ABC Inc. by Walt Disney Co., or AT&T’s acquisition of TCI corp., purchased by an exaggerated price, a common consequence in this type of “flashing” operations (Gaughan, 2007).

In the scope of the strategic versus opportunistic M&A deals’ issue, another reason that may drive a “thrill of the chase” may simply be the chance of making some fast money (S. Moeller & Brady, 2007). Indeed, M&A announcements may be trigged by market underpricing. As discussed earlier, behavioural M&A research argues that M&A waves occur as a result of temporary stock market misvaluations (e.g. Dong et al., 2006; Rhodes-Kropf et al., 2005; Rhodes-Kropf & Viswanathan, 2004; Shleifer & Vishny, 2003).

240

Therefore, one may argue that there are many reasons that can lead to “quick mergers”, both from theoretic and pragmatic viewpoints.

Finally, following Jensen & Ruback’s (1983) seminal paper, most literature using daily data was primarily concerned with measuring M&A returns for bidder and target firms’ shareholders, such returns being like a kind of proxy for stock prices movements. The ease and objectivity underlying the CAR-like methodology, made such studies very appealing for researchers, resulting in a very large number of publications. However, as mentioned before, such methodological simplicity is not shared by all literature, as several studies also use daily returns, though not merely to measure impacts on shareholders’ wealth. Indeed, as this type of earlier examination, based on pure returns, started to be exhausted, other strands of literature began pursuing different research objectives, leading to the development of the existing methodologies. As examples, it is possible to refer the studies of Ayers et al. (2002), or Norris & Ayres (2000), primarily concerned with the accounting method used, pooling versus purchase; studies concerned with the M&A market regulation, such as the one of Aktas et al. (2004); or studies about merger waves (e.g. Matsusaka, 1993). These studies used returns simply as means to examine several issues related to M&A activity. Also more noteworthy is the study of Branch et al. (2001), which used M&A activity daily data, with a multivariate model, to explore the M&A pattern of activity during the period 1982-1998. It is also noteworthy to refer to the specific “day-of-the-week” effects on the markets, which have been largely examined in literature (vid. e.g. Berument & Kiymaz, 2001; Galai & Kedar-Levy, 2005; M. Gibbons & Hess, 1981).

Accordingly, in order to test hypothesis one, four sets of completed M&A deals were prepared:

(i) Quarterly, 60 quarters, from 1994 to 2008; (ii) Monthly, 36 months, from 2000 to 2002; (iii) Weekly, 156 weeks, ibidem; and finally, (iv) Daily, 1096 days, idem.

241

Nevertheless, the weekend’s inclusion in the models using daily data posed two main questions: firstly, weekends are non-trading days. Their inclusion would therefore result in a mismatch with some independent variables being used in the model using daily data, such as stock prices indexes, for which only weekday data is available. Secondly, from a statistically point of view, weekend days represented undesirable and significant outliers and influential points, producing autocorrelation in the residuals.202 For these reasons, weekends were eliminated, with the total of daily observations consequently being reduced to 782 weekdays.203 204

For hypothesis two, another two sets of aggregated data of withdrawn M&A deals were arranged:

(v) Quarterly, 60 quarters, from 1994 to 2008; and (vi) Monthly, 36 months, from 2000 to 2002.

The average number of monthly observations itself is low, around twenty-five withdrawn deals. Consequently, it was not then feasible to test hypothesis two using daily or weekly aggregated data.

202 The 2000-2002’s study period comprise 314 weekend days, 157 Saturdays and 157 Sundays. Only two thirds of Saturdays had announcements (102), while Sundays were even quieter: 90 Sundays did not have any announcement. A total of 481 deals were announced during weekends, 275 on Saturdays and 206 on Sundays. The average number of announcements during a weekend day was only a deal and a half, while a weekday would produce in average almost twenty-five announcements. 203 As a consequence, the actual daily sample size was reduced from 19,758 to 19,227 completed deals. 204 In order to minimise the effect of weekends’ elimination, independence hypothesis tests for samples including weekends were conducted. No significant differences between means were found.

242

6.6 Conclusions

Following an examination of some issues related to existing M&A datasets, it has been found that SDC Platinum (Thomson Financial, 2006, 2009) was a suitable M&A data source, and therefore data from this provider was collected, being organised and structured in two main samples. Together with M&A data, were also presented other sources of economic and financial data. Altogether, this data will be used to test the research hypotheses, by means of quantitative analysis, as shown in the next chapter.

Different sets of data, using quarterly, monthly, weekly, and daily observations, were prepared in order to set the ground for testing the research hypotheses. Issues related to the use of very short-term data aggregations were also discussed, with particular emphasis to the examination of the use of daily data.

In conclusion, six sets of data are to be used for testing in chapter 7, using several models fitted according to the different hypotheses and datasets shown previously in this thesis.

243

Chapter 7 Models Development and Testing Results

7.1 Summary

This chapter is focused on the development of empirical research, which will serve to test the main research question. Besides a description of the metrics used, this chapter also covers a set of quantitative tests, supported by a statistical analysis of the results obtained. Therefore, it provides critical evidence for the discussion of the research hypotheses. Furthermore, the empirical research shown in this chapter offers additional evidence on M&A activity, and on its pattern, particularly for the period of study.

As discussed earlier, in chapter 4, the research models are designed to test the main hypotheses, and they are not particularly concerned about other possible sub-findings that such models’ results could provide. This is reflected not only in the approach to the research hypotheses, but also in the manner the empirical research was conducted, as shown throughout this chapter.

An example about the methodological view adopted follows. As shown later in this chapter, an estimation was made about the expected signs for the variables used in the regression models. Since the variables are to be tested, one could hypothesise ex ante as to whether a particular variable should be positively or negatively related with M&A activity, and draw conclusions from the results obtained. To illustrate this possibility, one can argue that it seems consensual that interest rates are negatively correlated with M&A activity. Indeed, if

244 interest rates are lowering, the M&A activity is supposed to increase, as financing M&A deals becomes cheaper. However, such type of findings is not relevant for the present research, at least when taken individually.205 Moreover, it is not a very long term study, covering several decades surrounding the event date, and, therefore, it may be acceptable to have variables in the regression models with ex post signs contradicting expected signs, i.e., ex ante. One cannot argue that such contradiction is entirely irrelevant, as a substantial number of ex post opposite variable signs could hint a model misspecification. But, as long as opposite sign results are accidental or justifiable, this possibility will not be regarded as an issue for this research, as it is not intended to draw any other conclusions beside the ones resulting from testing the main hypotheses.

Finally, following the examination of univariate descriptive statistics, the results obtained from testing the main research hypotheses are then subject to discussion, comprising several empirical tests in the context of bivariate and multivariate analysis, model validation, examination of outliers and influential points, and sensitivity analysis.

7.2 Introduction

There are a number of different methodologies, metrics, and techniques that could be used to test the research hypothesis. Nevertheless, literature suggests economic modelling as being a superior one, including the study of effects from economic policy, as proved by the widespread use of econometric models by policy makers (vid. e.g. Hendry & Mizon, 2005; McNown, 1986; Mizon, 2004). This is also suggested for M&A research. Pautler (2001), while referring to a specific M&A case, suggests the superiority of the use of econometric modelling over more simple testing procedures, such as a statistical comparison of group means.

205 The analysis of results of variables is important as a whole, not individually. This subject matter will be discussed in the final chapters of this thesis.

245

Similarly, Branch et al. (2001: 9), while developing a M&A “daily model”, also noted that “potential relationships can be explored in either a univariate or multivariate context”, concluding that it is preferable to choose “the greater completeness of a multivariate model”, based on two main reasons:

“First, a multivariate model removes the impact of the included variables allowing the variable-of-interest (in this case merger activity) to explain only what is otherwise unexplained. Second, somewhat similarly, an extensive multivariate model reduces the risk that the variable-of-interest is acting as a proxy for some other variable. Accordingly, we seek to isolate the overall market impact of merger activity, if any, in a particular framework.”.

This thesis pursues similar aims, although seeking to test primarily the impact of the new M&A accounting rules on M&A activity. Therefore, taking into consideration the objectives of this thesis, the use of multiple regression analysis in this thesis seems a valid option. However, the originality offered by the present research question implies the challenge of having to develop models with some degree of novelty. Indeed, despite the several models available in the literature, such as the case of Branch et al. (2001), they are not straightforwardly applicable in this thesis, as they were driven by different purposes. Nevertheless, it is a fact that many existing models do provide relevant contributions and were therefore incorporated in the models developed for this thesis. In fact, it can be argued that the models developed here assemble different parts from other models existing in M&A literature.

Besides Branch et al. (2001), which examined the pattern of M&A activity at daily level, in terms of noteworthy models found in the literature that provided major contributes for this thesis, are the time series models developed by Melicher et al. (1983), and several different autoregressive models, such as the likes of Barkoulas et al. (2001), Shughart II & Tollison (1984), and Town (1992). There are others that have also contributed to the models construction, but more indirectly, such as the case of the time series model of Golbe & White (1988), or the valuation model of Rhodes-Kropf & Viswanathan (2004). Seminal studies, such as the ones of Nelson (1959), and Markham (1955), have also been

246 considered, but to a very limited degree, mainly to have some additional justification, and supporting evidence during the variables construction stage. Together with other models previously mentioned in this thesis, they support the most relevant decisions made while developing this research. They will continue to be referred to throughout this chapter, whenever considered to be relevant, alongside with other models yet to be introduced in this thesis.

As discussed earlier in this thesis, whenever suitable comprehensive theories and models are missing, the use of a regression-based model may be recommendable, as it may work as a fine predictor.206 Nevertheless, in order to obtain a valid response from the values of the regressors, it is necessary to prepare a model carefully fitted from a sample large enough. Although models with a few observations appear to have more predictive power, since using small amounts of data means less possible abnormal circumstances introduced in the model, they are more likely to suffer from several methodological issues, which include the possibility of biased results. This is one of the reasons why, besides models based on quarterly, monthly, and weekly data, a model has also been prepared using daily data, since it provides more observations, therefore improving statistical interpretation, and reinforcing the accuracy of the parameter estimates.

The data aggregation used in the current study carries some issues related with the use of time series. Time series include cycle, seasonality, trend, and randomness. Cycles are usually reflected only in larger data aggregation periods,

206 The methodology adopted here to test the research hypotheses allows to make predictions on the M&A activity evolution. Indeed, any regression model has some intrinsic predictive power by default, regardless to be used or not, regardless being intentionally designed for such purpose or not. However, similarly to Branch et al. (2001: 9), whose modelling purposes were regarded as “explanatory rather than a predictive model”, such predictions are not relevant for achieving the research objectives in this thesis, in the sense that it is not important to make projections about the evolution of M&A activity beyond the periods 2000-2002, and 1994-2008. What is relevant is to examine the results within the sample periods. Therefore, such forecasting power will be solely used in order to test the goodness-of-fit of any constructed model, whenever justified.

247 such as quarterly or yearly ones. In the present study, specific patterns resembling somewhat a cycle-behaviour are to be treated with dummy variables.207 Seasonality is also to be treated with dummy variables. This leaves trend and randomness. The magnitude of randomness diminishes as the level of aggregation increases. Monthly data is less random than weekly data, as by averaging thirty days more randomness is eliminated than averaging only seven days. Conversely, as randomness decreases the trend included in data become more notorious. In daily data, randomness is more notorious while trend is absent or insignificant (Makridakis et al., 1998: 536). In this case, simple smoothing is preferred to other more complex procedures, such as Holts and Winters’s methods, as there are no particular concerns to capture trend effects. Nevertheless, it has been decided to use M&A data as raw as possible to minimise any misrepresentation. Instead of transforming raw M&A data, the use of dummy, adjustment, and lagged variables, was overall preferred in order to deal with trend and randomness.208 Any remaining trend is to be treated using polynomials.209

Following the information revolution, financial markets increased the chances of becoming more efficient, in the sense that more information could be more quickly priced (vid. e.g. Andersen et al., 2007; Gu, 2002; Gu & Finnerty, 2002).210 Greater efficiency means that markets behave increasingly like random walks. Makridakis et al. (1998) point out that this makes it impossible to

207 Dummy variable, or indicator variable, is a binary variable, which assumes value one or zero. There are several purposes for the use of dummy variables, such as to measure qualitative events, or to capture of diverse effects such as seasonality, holiday effects, and interventions (i.e. impacts), among others. 208 As examples of dependent variables transformations, vid. e.g. later equations (14) and (15). 209 Polynomials were used in the models using the 2000-2002’s sample as they provided a superior fitting. In the case of the 1994-2008’s sample, the logarithmic transformation of the dependent variable proved to provide a better fitting for the model used to test hypothesis one; while for the model used to test hypothesis two, a multiple regression model with linear time series was used, as the different fittings tested did not prove beneficial enough. 210 Nevertheless, the examination of markets’ efficiency evolution in recent decades does not provide consensual findings (vid. e.g. Gu & Finnerty, 2002).

248 predict the turning points using statistical methods. They also note that unpredictable, insignificant events could trigger turning points, just like the ‘butterfly effect’ in chaos theory, an extreme example, where it is suggested that the air displaced by a flying butterfly in a tropical forest can instigate a major hurricane a week or two later.211 Additionally, psychological effects are present in business and economics, and they have proved to be highly influential on the markets. Unpredicted sudden raises and crashes are often more related to human behaviour than to business and economic events, making analysts to label this type of behaviour as ‘irrational’.

If randomness dominates in a time series, it is possible, then, that a simple random walk model, or other naïve model, will have a predictive power similar to complex explicative models. This may not happen for all M&A markets worldwide, but it is more likely to be true for the US market, which is historically the most dynamic and efficient. Therefore, it is not surprising then that some literature claims that random walk hypothesis describes better the M&A activity (e.g. Chowdhury, 1993; Shughart II & Tollison, 1984), although a number of authors disagree, particularly some who have confirmed the existence of M&A waves (e.g. Golbe & White, 1993; Town, 1992). As described below and in the following section, the present research considers the random walk hypothesis. Nevertheless, due to its specific purposes, it refers to both strands of literature referred above, regardless the viewpoints about random walk hypothesis, as this thesis shares the use of similar methodologies, including the same type of data aggregations and variables. The M&A market certainly has a lower level of efficiency when compared with stock markets - or at least M&A activity tends to be more volatile than some other macroeconomic time series (Blair et al., 1991). Nevertheless, M&A and stock exchange markets are closely bonded and they do share many common characteristics. Moreover, these

211 Chaos theory is authored by Edward Lorenz, a meteorologist which pioneered studies of atmospheric dynamics in the early 1960’s. The “butterfly effect” concept was introduced in 1972, when Lorenz presented a seminal paper before the American Academy for the Advancement of Science in New York, entitled Predictability: does the flap of a butterfly's wings in Brazil set off a tornado in Texas?

249 characteristics become more visible whenever data aggregation is lower, which is the case in the models using the 2000-2002’s sample.

The use of a low level of data aggregation leads to an additional issue, concerning the diversity of exogenous explicative factors that can be employed. The number of different types of daily, weekly, and monthly data available and feasible to relate with M&A activity is more limited, which therefore reduces the number of explicative variables that can possibly be considered. For example, GDP data is only available quarterly and the adoption of extrapolation techniques may not be entirely trustworthy. To mitigate the impact of this constraint on the development of the models, the pattern of the M&A activity has been researched in depth, resulting in a relatively higher weight of endogenous explicative factors, as less exogenous variables were available, particularly for the models using daily and weekly data. Nevertheless, as shown throughout this chapter, every possible variable was used whenever possible. It was the case of GDP variable, which has been included not only in the model using quarterly data, but also in the one using monthly data, ensuring this explicative factor to be subject to testing whenever feasible.

Finally, many model-selection methods are available to help with the modelling process. These possibilities include: methods to select models with the highest value of R2, or highest values of adjusted R2; stepwise regression; or other measures such as Mallow’s Cp statistic, Akaike’s Information Criterion (AIC) statistic, and Schwarz Bayesian Information Criterion (BIC), amongst others. These procedures and many other model-selection methods are widely reviewed in the literature (see e.g. Brockwell & Davis, 1996; Draper & Smith, 1981; Hocking, 1976; Judge et al., 1988).

Stepwise regression is a method that makes it possible to select the relevant explanatory variables from a set of candidate variables. This specification procedure includes different approaches, such as stepwise forward regression, or stepwise backward regression (see e.g. Draper & Smith, 1981). The stepwise forward regression method begins with no variables in the model, and then

250 starts adding statistically significant variables, while the stepwise backward regression method begins with all variables in the model and then starts eliminating lesser significant variables. Both methods have several variations, being its use common in M&A research (e.g. Covin et al., 1997; Pfeffer & Salancik, 2003; C. Yang & Qiu-sheng, 2006).212

The use of stepwise regression in this thesis is justified by several main reasons. First, as a consequence of a relatively limited number of explanatory variables available, similar variables were included in the long list of variables that could possibly figure in the final models, being therefore needed to select the most significant variables, in order to avoid multicollinearity. Another justification is that this possibly improves the quality of the models fitting, as stepwise regression considers every variable for the testing procedure, but only includes statistically significant dependent variables in the final versions of the models, therefore increasing the possibility of obtaining models properly validated according to the assumptions of regression analysis. Finally, the use of stepwise regression is also related with the main point of this research: to test whether the ‘event’ variables have any predictive value in the models. If they do not, then it will mean that the effect of the accounting changes on the M&A activity was not statistically significant.

Among the diverse stepwise approaches available, it has been selected the backward elimination.213 With the backward elimination procedure, all variables from the long list, which includes the ‘event’ variables, are initially considered and then subject to potential elimination in case of low statistical significance. In the statistical software package SAS 9.1, the backward elimination procedure starts by calculating the F statistics for a model, including every independent

212 Stepwise regression should not to be confused with the concept of “stepwise” or “step-wise” procedure, a managerial technique used in M&A due diligence, implementation, and integration. 213 As any methodology or technique, every statistic has its own drawbacks. For stepwise regression drawbacks see e.g. Stevens (2002).

251 variable.214 Then the variables are deleted from the model one by one until every variable remaining in the model produces F statistics significant at the level specified by the user (0.10 level by default). At each step, the variable presenting the smallest contribution to the model is deleted.215 SAS’s procedure is also similarly followed by other statistical software packages (e.g. StatPoint Inc., 2006).

In summary, several models have been developed in order to test the research hypotheses, but under some constraints, such as regarding the number of explanatory variables available to be tested. Considering time as a factor, the models development combined multiple regression with time series. Moreover, the backward selection procedure has been employed, not only with the purpose to assess the potential significance of the event variables; but also to enhance the fitting and validation of the models. This order of priorities is justified by the main objective of the present study, which is to assess the possible effects on M&A activity from the M&A accounting changes, rather than to study in depth other explicative factors, or the trends related to the M&A activity itself. Nevertheless, because a model with enough predictive power is a sine qua non condition for validating its outcomes, none of the factors concerning the M&A activity can be simply disregarded, as shown by the examination made in this chapter about the patterns and behaviour of M&A activity.

7.3 Construction of metrics

The models of this thesis have been designed in order to capture the impact of the M&A accounting changes on the number of announced deals and on the level of withdrawn deals. Accordingly, multiple regression with time series was adopted, together with the use of the stepwise procedure as a methodology of

214 Statistical Analysis System, henceforth referred as SAS 9.1, is a software package copyright of SAS Institute, Inc. 215 Described procedure adapted from SAS 9.1 “Help and Documentation” (SAS Institute Inc., 2003, 2004).

252 selection of explanatory variables, as described in the literature by authors such as Makridakis et al. (1998).

The models combine the characteristics of a generic explanatory model, in what concerns the explanatory variables, with time series, where there is an attempt to capture the importance of the time factor over time. Random walk hypothesis is assumed and it is used to address the issue of the non-stationary data. A brief methodological review, broadly supported in Makridakis et al. (1998), but also in other literature as to be referred to, follows. time series: The models used in this thesis embrace time series analysis. A time series consists of an ordered sequence of data observed at time intervals often equally spaced. Time series analysis tries to understand the patterns underlying a given time series. It can also be used to make predictions. Time series analysis has been largely used in M&A research, particularly whenever examining the M&A pattern of activity and also its wave pattern of occurrence (see e.g. Barkoulas et al., 2001; Chowdhury, 1993; Golbe & White, 1988, 1993; Guerard, 1985; Melicher et al., 1983).

Unlike cross-sectional regression, where all data is measured at the same time, in a time series regression, data is measured over time. A time series model is a function that relates the value of a time series to previous values of that time series, its prediction errors, or other related time series (Makridakis et al., 1998: 616). Time series data can be decomposed in pattern plus error, which is equivalent to data = f (trend-cycle, seasonality, error). A simple notation such as (1) describes the series X which is indexed by Xn natural numbers.

X = {XX12, ,..., Xn} (1) random walk: As mentioned before, the nature of M&A pattern and the data aggregations used in this research make it reasonable to assume random walk hypothesis

253

(vid. also Chowdhury, 1993; Shughart II & Tollison, 1984). The observation of previously exhibited Fig. 4.1 also helps to corroborate this assumption. The M&A activity presents upward and downward trends that can reverse suddenly, often unpredictably. This pattern resembles stock markets indexes performance and random walk series behaviour. Additionally, the pattern shown in Fig. 4.1 suggests that M&A series in the period of analysis is stationary in the variance, but, conversely, is non-stationary in the mean. The assumption of random walk hypothesis allows to address the issue of the non-stationary data.

A random walk model, for time period t, can be shown as

YYtt=+t−1 ε (2) where:

Yt is the value of the series for time period t,

Yt-1 is the value of the series for time period t-1, and

εt is white noise. autoregression: Autoregression is a type of regression, but instead of the dependent variable being related to other explanatory variables, it is related to past values of itself at varying time lags (Makridakis et al., 1998: 591). Different types of autoregressive models have been used in M&A literature (vid. e,g. Barkoulas et al., 2001; Melicher et al., 1983; Shughart II & Tollison, 1984; Town, 1992). An autoregressive (AR) model has the form

YbbYbYttpt=+01122tt−− + ++... bYek − +. (3)

Time series models can assume many forms, such as autoregressive (AR) models and moving average (MA) models. Combinations, such as autoregressive moving average (ARMA), and autoregressive integrated moving average (ARIMA) models, were made popular by Box & Jenkins (1970) and are still widely used in

254 the literature, as other variants as well. An classical ARIMA(p, d, q) model can be decomposed by (Makridakis et al., 1998: 336): AR: p = order of autoregressive part; I: d = degree of first differencing involved;216 MA: q = order of the moving average part.

An autoregressive model of order one can be classified as either an AR(1), or an ARIMA(1,0,0), because there is not any difference involved (I), or MA. From equation (3), if observation Yt depends on Yt-1, and if the value of the autoregressive coefficient φ1 is restricted to fluctuate over the interval from -1 to +1, an ARIMA(1,0,0), or AR (1), equation can be written as

YYett=+ω ϕ11t− + (4)

If φ1 = 0, Yt is equivalent to a white noise series. White noise is when there is no pattern or whatsoever in the time series. A white noise model can be classified as an ARIMA(0,0,0), because there is not any AR, no difference involved (I), or MA

Yett=+ω . (5)

If φ1 = 1, Yt is equivalent to a random walk series

YYett=+ω t−1 +. (6)

In this thesis, the models are constructed following a basic first-order autoregressive model, similarly to Shughart II & Tollison (1984).

216 Differenced series is the change between each observation in the original series

' YYYttt=−−1 .

The use of differenced series results in n-1 observations since it is not possible to compute a difference Yt’ for the first observation.

255 multiple linear regression: A multiple linear regression equation, for a model with k explanatory variables, and for time period t, can be represented as

YXXXtt=+β011,22,ββtt + ++... βkkt, + ε (7) where:

Yt is the dependent variable (also called response variable, or criterion) at time t,

Xk,t are the k independent variables (also known as explanatory, or predictor variables) at time t,

βk are the k regression coefficients (unknown parameters, to be estimated), and

εt is the error term, a random variable, at time t.

Multiple regression analysis can be found largely in M&A research, but as most relationships are non-linear, several linearisation procedures are employed. Linearization forms used in this thesis are shown below. non-linear relationships: Frequently, the relationship between the response variable and the explanatory variables is not a linear one. Indeed, non-linear functions often explain better these relationships. Besides the linear relationship, other forms of functional relationships connecting the response variable and the explanatory variables are used to fit models, such as polynomials and exponentials.

Polynomial functions are linear in the parameters and can be presented as follows

2 j YXXXtttjtt=+α ββ12 +++... β + ε ,j = 1,...,k (8)

256

th where the independent variable Xt is taken from the first power up to the j power. The equation above, i.e. polynomial functions of k order, can be synthesised as

k j YXtjtt=+αβ∑ +ε . (9) j=1

If the model consists only of Xt taken to the first power, then it will be a simple linear regression model, or a first-order polynomial, also referred to as a first-

-degree polynomial (Lomax, 2007: 407). A second-order polynomial includes Xt taken to the second power, and is commonly referred to as a quadratic model, i.e. a curve with one bend in it rather than a straight line, like in linear models. A quadratic functional relationship, or a second-order polynomial, can be written as

2 YXXtttt=+α ββ12 + + ε. (10)

Likewise the equivalence of equations (8) and (9), equation (10) can be shown to be equivalent to

2 j YXtjtt=+αβ∑ +ε (11) j=1

A cubic relationship, or a polynomial of order three, includes X taken to the third power, i.e. a curve with two bends in it. Likewise shown in previous transformations, a third-order polynomial can be written as

3 j YXtjtt=+αβ∑ +ε . (12) j=1

Further information on polynomial regression models can be obtained in diverse literature, such as in Weisberg (1985), Seber & Wild (1989), or Bates & Watts (1988). Polynomials are normally used in M&A literature with the maximum polynomial order of three (vid. e.g. Barkoulas et al., 2001).

257

Finally, unlike linear relationships and polynomials, an exponential function is non-linear in the parameters

α++ββ11,XXtt 2 2, +++... βkkt X, εt Yet = . (13) linear transformations using logarithms: Many non-linear functions can be transformed into linear functions. This can be done using logarithms to form linear models. An example of a linear transformation of a non-linear relationship using logarithms is given by the transformation of the dependent variable Yt, onto logZt

p ZLogtt=++()α βε11, Xt . (14)

The linearisation of an exponential function, as exhibited in (13), is also possible with a logarithmic transformation of the dependent variable Yt onto logeZt:

ZLogte=+++++(...)α ββ11, Xtt 2 X 2, βkkt X, ε t. (15)

Several non-linear relationships and linear transformations have been employed in the initial stage of the development of the models. Nevertheless, it has been found that polynomials would provide, in general, better model fittings, given their superior balance between a higher predictive power, and the fulfilment of the required conditions for models validation. There is, however, an exception in the case of the model developed to test hypothesis one using quarterly data, as it has been found that a logarithmic transformation of the dependent variable resulted in a better fitting rather than using polynomials.

To resume, in order to support the testing aims of this thesis, the construction of the models for this research involved assembling different contributions from several models existing in the literature. This process has resulted in the development of models assuming the random walk hypothesis; combining multiple regression with time series analysis; together with the use of first-order

258 differences; employing polynomials, or a logarithmic transformation of the dependent variable; and, finally, that use stepwise regression for the selection of the most significant variables.

7.4 Statistical models for hypotheses testing models developed to test hypothesis one: A set of three models, using monthly, weekly, and daily data, were prepared using the same foundations in order to examine the association between M&A activity and a large set of variables, from which the event ones assume a particular interest. The main model, hereafter called the basic model, for time period t, and with βj, δl, λi, ζp, and ξ, regression coefficients, can be outlined as

32mn j MAtjt=+αβ∑∑ Per + δllt Exog, + ∑ λ iitppt Endog,,+ ∑ ζ Event + jl==11 i = 1 p = 1 (16)

++ξεMAt−1 t where:

MAt is the number of M&A deals announced for time period t, j Pert is the period, or time variable, with j = 1 if linear, and with j = 2, 3, if quadratic, or cubic, respectively,

Exogl,t are the m exogenous explanatory variables, such as stock market prices, or other economic factors,

Endogi,t are the n endogenous explanatory variables, related to M&A activity and seasonality,

Eventp,t are two dummy variables developed to control the impacts on M&A activity from FASB’s changes in M&A accounting,

MAt-1 is a lagged variable, which lags the dependent variable MAt by one period, and

εt is the error term.

259

As mentioned before, in the case of the model developed to test hypothesis one using quarterly data, the logarithmic transformation of the dependent variable resulted in a better fitting than with the use of polynomials. Therefore, an alternative basic model, supporting a fourth model using quarterly data, has been accordingly fitted as shown below

mn 2 Log MAtiitppt=+αδ∑∑llt Exog, + λ Endog,,+ ∑ ζ Event + li==11 p = 1 (17)

++ξεLog MAt−1 t where:

Log MAt is the logarithm of the quarterly number of M&A deals announced, and

Log MAt-1 is a lagged variable, which lags by one period the dependent

variable Log MAt , i.e. the logarithm of the number of M&A deals announced during a quarter.

The alternative basic model, shown in equation (17), resulted in the construction of a model using quarterly data

322 Log MAtiit=+αδ∑∑∑Quarter ,, + λllt SP500 , + ζ mmtt Fed + ω IP + ilm===111 (18)

+++++φ MKTCttttγϕθξ GDP TD Event Log MAt−1 + ε t where:

Quarteri,t are three dummy variables representing the quarters of a year, with i = 1 to 3, representing the first three quarters of the year, respectively,

SP500l,t are two stock prices index variables, measured by average and closing values of a quarter,

Fedm,t are two interest rates variables, measured by average and last values of a quarter,

IPt is a quarterly industrial production variable,

MKTCt is a quarterly market capitalisation variable,

260

GDPt is a quarterly GDP variable,

TDt is a variable that accounts for the number of trading days during a quarter, and

Eventt is a dummy variable created to capture the potential effect from FASB’s standards.

From the basic model, having equation (16) as a proxy, the model using monthly data can be specified as

3112 2 j MAtjtiitmmtt=+αβ∑∑ Per + δ Month,, + ∑ λllt SP500 , + ∑ ζ Fed + ω IP + ji==11 l = 1 m = 1 (19)

++++++φγϕθξεMKTCtttt GDP TD Event MAt−1 t where:

MAt is the monthly number of M&A deals announced,

Monthi,t are eleven dummy variables representing the months of a year, with i = 1 to 11, representing the months from March to January, respectively,

SP500l,t are two stock prices index variables, measured by average and closing values of a month,

Fedm,t are two interest rates variables, measured by average and last values of a month,

IPt is a monthly industrial production variable,

MKTCt is a monthly market capitalisation variable, and

TDt is a variable that accounts for the number of trading days during a month, while a model using weekly data can be presented as

34 2 2 j MAtjtiitmmt=+αβ∑∑ Per + δ Week,, + ∑ λllt SP500 , + ∑ ζ Fed + ji==11 l = 1 m = 1 (20)

++++φ E__ BoMttϕθ TD Event t ψ Event ED t ++ ξ MAt−1 ε t

261 where:

MAt is the weekly number of M&A deals announced,

Weeki,t are four dummy variables representing the weeks in a month, with i = 1 to 4, representing the first four weeks in a month,

SP500l,t are two stock prices index variables, measured by average and closing values of a week,

Fedm,t are two interest rates variables, measured by average and last values of a week,

E_BoMt is a dummy variable added to capture calendar and seasonal effects,

TDt is a variable that accounts for the number of trading days during a week, and

Event_EDt is a dummy variable constructed to capture the potential effect from the ED that preceded FASB’s standards.

Finally, a model using daily data can also be estimated as

34 j MAtjtiitttt=+αβ∑∑ Per + δ Weekday, + λ SP500 + ζ Fed + ϕ Hol + ji==11 (21)

+++++ϖ HS__ Exttttt+φθψξε E BoM Event Event _ ED MAt−1 t where:

MAt is the daily number of M&A deals announced,

Weekdayi,t are four dummy variables representing the weekdays, with i = 1 to 4, representing Monday to Thursday, respectively,

SP500t is a stock prices index variable, measured by daily closing values,

Fedt is a daily interest rates variable,

Holt is a dummy variable which signals a non-trading day, and

HS_Extt is a dummy variable that captures the impact of reduced trading days, holiday seasons, and extraordinary events.

262 withdrawn M&A models utilised to test hypothesis two: Two sets of models, using quarterly and monthly data, have been developed to test hypothesis two. They are outlined similarly to the basic model used to test hypothesis one, as shown before in equation (16). However, as a dependent variable, the models constructed to test hypothesis two use the number of withdrawn M&A deals, instead of the number of announced deals, which, in turn, have been added as an explanatory variable.

Equation (22) replicates the announced M&A deals main model, as shown in equation (16), but does not use polynomials, as it represents a multiple regression model with linear time series

322 WITHttiitmmt=+αβ Per +∑∑∑ δ Quarter,, + λllt SP500 , + ζ Fed +ω IPt + ilm===111 (22)

++φ MKTCt γ GDPtt++ j TDρψ MA t + Event t + ξ WITHt-1 + ε t where:

WITHt is the quarterly number of M&A deals withdrawn,

WITHt-1 is a lagged variable, which lags the dependent variable WITHt by one period, and

MAt is the quarterly number of M&A deals announced, while equation (23) replicates the monthly announced M&A deals model, as exhibited in equation (19), but regresses the number of withdrawn M&A deals instead

3112 2 j WITHtjtiitmmt=+αβ∑∑ Per + δ Month,, + ∑ λllt SP500 , + ∑ ζ Fed + ji==11 l = 1 m = 1 (23) +++++++φγωϕρψMKTCttttt GDP IP TD MA Event t

++ξεWITHt−1 t where:

WITHt is the monthly number of M&A deals withdrawn.

263

7.5 Variable definitions and predictions

The process of constructing the variables is drawn largely on the literature on M&A and on the analysis of the M&A pattern during the period of study. This section starts by exhibiting the long list of exogenous explanatory variables constructed, following by the examination of time-related factors and endogenous explanatory variables, such as time, seasonality, trading day variation, and holiday effects; and finishes with the examination of the intervention variables, together with a reference to the use of lagged variables in the models. dependent variables: In the present research, the number of announced M&A deals is a proxy for M&A activity, and its pattern, being represented by the dependent variables MA, or Log MA, a logarithmic transformation of MA; while the number of M&A withdrawn deals is represented by the dependent variable WITH. Models using MA and Log MA variables have been conceived to test hypothesis one, while the models using WITH variable have been designed to test hypothesis two. exogenous explanatory variables: As mentioned earlier, several factors are likely to contribute to the understanding of the pattern of M&A activity. Movements on stock markets prices, interest rates, GDP, and industrial production are exogenous explanatory factors for which there are reliable datasets available and were therefore used in the models whenever feasible (see e.g. Becketti, 1986; Golbe & White, 1988; Melicher et al., 1983; Weston et al., 1990). Information about the long list of exogenous explanatory variables that will be subject to the stepwise backward elimination is as follows.

264

The S&P 500 Composite index has been selected as a proxy for stock prices indexes. More precisely, has been utilised the S&P 500 Composite – default datatype (PI), which is the default Datastream data type for equity indices.217 The adoption of this index is justified by its widespread use in literature as a proxy for stock prices index evolution in the USA. Additionally, the use of S&P 500 in this research is also justified since this index has also been previously used in this thesis as a criterion of selection for both survey and annual reports analyses. As a proxy for interest rates, it has been selected the US Federal Funds (effective) – Middle Rate.218

In the case of the models using quarterly, monthly, and weekly data, two different views were considered for the variables using stock prices and interest rates. These approaches arose from the possibility of choosing from closing values of the last trading days of the quarter, month, and week, versus using quarterly, monthly, and weekly average values. To avoid any misjudgement, two types of variables were constructed for stock prices and interest rates: one

217 According to Datastream, the index is calculated as follows:

I0 = index value at base date = 100 n ∑()PNtt∗ 1 IItt=∗−1 n ∑()PNftt−1 ∗∗ 1

Where:

It = index value at day t

It-1 = index value on previous working day (of t)

Pt = unadjusted share price on day t

Pt-1 = unadjusted share price on previous working day (of t)

Nt = number of shares in issue on day t f = adjustment factor for a capital action occurring on day t n = number of constituents in index The summations are performed on the constituents as they exist on day t. 218 According to Datastream, the federal funds rate is the interest rate at which depository institutions lend balances at the Federal Reserve to other depository institutions overnight. The daily effective federal funds rate is a weighted average of rates on trades through New York brokers. Rates are annualised using a 360-day year or bank interest.

265 uses closing values, while the other one uses quarterly, monthly, or weekly average values. The two types were therefore included in the initial models, being the selection of the most adequate variables entrusted to the stepwise backward elimination regression procedure.

Other three explanatory variables were employed, but only on the models using quarterly and monthly data:

(i) industrial production, more precisely, the US Federal Reserve Board’s industrial production index, which measures the real output of manufacturing, mining, and electric and gas utilities industries. The data provided is seasonally adjusted, and it is available monthly;

(ii) GDP, also seasonally adjusted at annual rates, but with estimates available only quarterly. For the models using monthly data, a monthly interpolation was initially considered, but later withdrawn, since this procedure is not reliable, as experts in general recognise. Nevertheless, the Business Cycle Dating Committee of the National Bureau of Economic Research (NBER), uses monthly estimates to measure cyclical peaks and troughs. However, such peaks and troughs, whenever measured by GDP and GDI (Gross Domestic Income), do not always coincide with the Bureau of Economic Analysis (BEA), an agency of the US Department of Commerce, which is one of the world’s leading statistical agencies (vid. e.g. Bureau of Economic Analysis, 2009; Fixler & Grimm, 2008). Furthermore, although BEA publishes quarterly estimates, it does not publish monthly GDP or GDI estimates (Grimm, 2005). Neither does NBER. Instead, NBER uses monthly estimates prepared by Macroeconomic Advisers, a consulting firm, which cannot be considered as authoritative by any means.219

219 According to the National Bureau of Economic Research (2009):

“Many of the ingredients of the quarterly GDP figures are published at a monthly frequency by the Bureau of Economic Analysis. Macroeconomic Advisers aggregates them, and then uses a statistical procedure to adjust the monthly estimates for each quarter to make them consistent with the Commerce

266

Therefore, in monthly data models this variable was kept constant during the term following the latest quarterly GDP value available; and, finally,

(iii) together with the stock prices index variable, a market capitalisation variable was also included. This variable is made from the sum of quarterly, or monthly, market capitalisation of all companies listed at stock markets in the USA: NYSE, NASDAQ, and AMEX.220

In terms of variables predictions, there is a broad consensus within the literature about a positive relationship between M&A activity and movements on stock prices (e.g. Beckenstein, 1979; Guerard, 1985; Markham, 1955; R. L. Nelson, 1959; Steiner, 1975), and between M&A activity and business cycle/GDP (e.g. Golbe & White, 1988; R. L. Nelson, 1959; Steiner, 1975; Weston et al., 1990). For industrial production, the majority of literature found a positive relationship with M&A activity (e.g. Gort, 1969; Markham, 1955; Mitchell & Mulherin, 1996), but some found that relationship to be weak (Melicher et al., 1983; R. L. Nelson, 1959), to non-existent (Guerard, 1985; Weston, 1953, 1961). For interest rates, the majority of studies observed a negative relationship (e.g. Becketti, 1986; Golbe & White, 1988; Melicher et al., 1983; Weston et al., 1990), but conversely some authors found a positive relationship (e.g. Beckenstein, 1979), even if a non-significant one (Steiner, 1975).

In respect to the expected signs, positive signs are expected from stock market indexes and capitalisation variables, as well from GDP and industrial production variables. For interest rates, the literature findings are not unanimous, therefore one can admit both positive and negative signs. In this

Department’s official quarterly figure. The monthly GDP numbers are fairly noisy and are subject to considerable revision”.

220 The following abbreviations are used in the models: SP500 for stock prices index measured by closing values, and SP500 Av if measured by average values, MKTC for market capitalisation, Fed for interest rates measured by end of period values and Fed Av if measured by average values, IP for industrial production, and GDP for gross domestic production.

267 thesis, for the period 1994-2008, a negative sign would be more expectable, because a decrease on interest rates should theoretically favour M&A activity, as debt becomes more attractive to finance deals. Nevertheless, between 2000 and 2002, the interest rates suffered several major and swift cuts, and the M&A activity decreased. Consequently, during 2000-2002, the M&A activity and interest rates are more likely to be positively related. time and endogenous explanatory variables: It has been previously discussed that endogenous factors play an important role in the study of M&A activity (see e.g. Branch et al., 2001; Golbe & White, 1988, 1993; Town, 1992). Similarly, it has also been mentioned that regression models often include linear and higher order polynomials (Makridakis et al., 1998: 610), not being advisable to exceed a cubic relationship (Barkoulas et al., 2001). Regarded as essential to capture the evolution through time of the M&A pattern of activity, time variables were included in the long list of variables. Accordingly, a time variable called period, or Per, was added, assuming the serial number of the observation. Together with Per_2, and Per_3, which equal Per squared and Per cubic, respectively, the formation of polynomial relationships becomes a possibility.221 Indeed, the selection of Per variable by any model would result in the inclusion of a linear time trend in the regression models, while the selection of Per_2 and Per_3 would involve polynomials of order two and three, respectively. Other polynomials of order superior to three were not used to prevent any possible over-fitting of the models, in line with Barkoulas et al. (2001). Only the model employing quarterly data used to test hypothesis one did not include Per variables, as a logarithmic transformation was used instead, since it provided a superior fitting. Because the number of

221 The statistical software packages used in this thesis compute polynomial regressions automatically, but with limitations in terms of options and analysis available, being the most significant the impossibility of combining polynomial regression with stepwise regression. To overcome such limitation, stepwise regression was selected and computed automatically together with the inclusion of numeric time series Per variables. More precisely, the period variables were manually introduced in order to have the algebraic equivalent to polynomials, as if computed automatically, being tested together with the remaining explanatory variables using stepwise regressions (vid. e.g. StatPoint Inc., 2005, 2006).

268 announced and withdrawn M&A deals decreased in the period 2000-2002, and as the number of withdrawn deals has also decreased during the period 1994- -2008, negative values are expected for Per variables.

An in depth analysis of the movements of M&A activity on a daily basis during the period 2000-2002, based on data collected for this thesis (Thomson Financial, 2006), makes it possible to conclude that stock market’s calendar proves to be influential, given that:222

(i) announcements are unusual during non-trading days. As mentioned before, in Chapter 4, weekends and holidays are particularly poor in announcements, often with none or with only a single record;223

(ii) reduced trading days affect the M&A activity in a negative way. This negative impact may be reinforced in the case of four-day weekends, when trading floors close early on the Monday preceding the holiday placed on a Tuesday; or, on the other hand, when the market closes early on the Friday following a holiday placed on a Thursday;

(iii) holiday seasons also affect negatively the M&A activity. It is the case of Christmas season, which has at least two non-trading days - Christmas Day and New Year’s Day - and a possible half-day trading session if Christmas Eve is

222 The analysis made from evidence collected in this thesis is often supported by existing literature, as to be referred to throughout this chapter. Official information about stock markets closures was taken from NYSE’s website (New York Stock Exchange, 2006a, 2006b). 223 In addition to the analysis made earlier, it is important to underline that the variance on the number of announcements is high during non-trading days. The reduction of the number of announcements is generally lower on holidays than at weekends, although specific holidays, such as Thanksgiving Day and Christmas, record a very low activity, being usually zero on Christmas Day. Regarding weekends, announcements are more likely to occur on Saturdays than on Sundays. From a total of 157 weekends, i.e. 314 Saturdays and Sundays, during the period 2000-2002, only 169 days, 102 Saturdays and 67 Sundays, had announced deals.

269 placed on a weekday. Other possible market special closures may occur in holiday seasons (e.g. a Friday after a Christmas Day placed on a Thursday);

(iv) a concentration of announcements is likely to occur following a holiday placed at the beginning of the week, a long weekend, or a holiday season period. In opposition, that concentration is likely to be brought forward in anticipation of flat calendar periods. For example, a Friday is likely to be busier in M&A announcements if it precedes a long weekend or a holiday season period. Another example is the paradigmatic case of the Thanksgiving Day, which is celebrated on the Thursdays of the last full week of November – in this case M&A activity tends to be anticipated before this very traditional long weekend; and, finally,

(v) unpredictable events may affect the normal markets operation and the M&A activity. It was the case of the terrorist attack on the World Trade Center (WTC). Following the attack, the New York stock markets were closed from 11 to 14 September 2001. One year later, on 11 September 2002, the NYSE opening was delayed until 12:00 noon out of respect for the memorial events commemorating the first anniversary of the attack on the WTC.

To handle the issues brought by the stock market’s calendar and events, several actions were taken and new explanatory variables were added to the models. Weekends were removed from the model using daily data, while holidays were kept, but treated instead with a dummy variable Hol, which takes value one, if a holiday is a non-trading day, or zero, otherwise. Another dummy variable, HS_Ext, was added to the model using daily data to account for the effect of reduced trading days, holiday seasons, and extraordinary events. Finally, an adjustment variable, number of trading days, or TD, was added to the models using quarterly, monthly, and weekly data. This categorical variable accounts for the total number of trading sessions during a month or a week. An ordinary trading day accounts for one, while a half-day trading session only totals 0.5.224

224 Instead of using a dummy variable, the adjustment could be done directly in the dependent variables, dividing the number of M&A deals, or withdrawn deals in case of hypothesis two, by

270

In terms of expected signs, negative signs are expected for Hol and HS_Ext variables, since they affect negatively the M&A activity, while a positive sign is expected for TD, since the number of M&A deals announced and withdrawn is likely to be positively related to the total number of trading days during a week, a month, or a quarter.

Stock markets and M&A activity share interesting seasonal patterns, such as:

(vi) a concentration of announcements is likely to occur on the first days of the month.225 This tendency to peak may be reinforced whenever a new quarter begins. These patterns are consistent with the ‘first-trading-day-of-the-month phenomenon’, which consists in a tendency for higher movements in the US stock markets on the first days of the month (Hirsch, 2004: 62) 226, and with the finding that in recent years the first month of a quarter is the most bullish in Dow Jones industrials and S&P 500 (Hirsch, 2004: 74).227 Typically, the peak of M&A deals occurs on the first trading day of the month. If the first day of the month is a non-trading day, or if a holiday is placed on the first days of the month, the peak may be then brought forward to the last days of the previous month, or may be split between the last and the first days of the month. Since

the number of trading sessions. However, the trading day dummy variable provided better results than the ones obtained by the adjustment made of the dependent variables. Additionally, the use of a dummy variable has the advantage of avoiding the transformation of the depending variables, keeping therefore M&A data raw in the testing models. 225 As suggested by evidence collected for the 2000-2002’s sample (Thomson Financial, 2006). During this period, the average number of announcements in the first week of the month exceeded the remaining weeks by around 20%. 226 For example: on the first day of January 2000, a Saturday, 17 deals were announced. This number of announcements is abnormal for a Saturday and can only be justified by a coincidence of positive effects such as the ‘end-and-beginning-of-the-month phenomenon’, and the beginning of a new quarter which is, cumulatively, the beginning of a new year. On the day after, a Sunday, the M&A activity returned to normal, since not even a single deal was announced, which is a normal condition for a weekend day. 227 According to Hirsch (2004: 62), from 2 September 1997, to 1 July 2004, the Dow Jones index gained 2711.74 points. During this period, the first day’s of each month accounted for a total 3559.06 Dow points, while the remaining 1635 days recorded a total 847.32 negative points.

271 announcements may occur in both non-trading and trading days, it seems therefore more appropriate to label this positive effect on M&A activity as an ‘end-and-beginning-of-the-month phenomenon’, since the change of month may take place during a non-trading day, be it a weekend, a holiday, or as a consequence of an extraordinary event.228 This “turn-of-the-month” pattern can also be found in financial markets and is well documented in behavioural finance literature (e.g. Ariel, 1987; Dickinson & Peterson, 1995; Lakonishok & Smidt, 1988). Indeed, while testing ninety years of daily data on the Dow Jones index, Lakonishok & Smidt (1988: 403) found “evidence of persistently anomalous returns around the turn of the week, around the turn of the month, around the turn of the year, and around holidays”. Abnormal returns were found for eight days around the turn of the month, with a particularly high concentration of abnormal movements in the last and in the first three trading days of the month, being this particular period labelled by the authors as a “strong turn of the month effect” (Lakonishok & Smidt, 1988: 417); and,

(vii) the majority of announcements occur in the beginning of the week, while Friday is the weekday with the fewest announcements. This behaviour is also in line with a more recent pattern in stock markets. According to Hirsch (2004: 132), based on S&P 500 index, Monday was the worst trading day of the week, in opposition to Friday, between 1952 and 1989. However, a reversal occurred in 1990, when Monday became consistently the best day of the week for the Dow Jones index, except for 2001 and 2002. Additionally, from 1992 to 2004, the bulk of Dow Jones index gains were made on the first two days of week, while Friday was the worst weekday. This “Monday effect” was also found in the M&A activity. Branch et al. (2001), examined and tested the concentration of M&A announcements on weekdays, using different sets of SDC Platinum data, resulting in evidence supporting the hypothesis that the number of M&A announcements tended to be higher on Mondays, for the period from 1982 to 1998. The analysis of the sample used in this thesis, corroborates the existence

228 During the period 2000-2002, the average number of announcements in the week that included the turn of the month, outperformed the remaining calendar weeks by an average of 21.2% (Thomson Financial, 2006).

272 of this effect, in this case for the period of 2000-2002, as the average number of announcements on Mondays exceeded the average number of announcements during the remaining working days by 21.77% (4,212 announcements in 143 days versus 14,876 announcements in the remaining 615 working days).

Furthermore, according to Branch et al. (2001: 6), Tuesday is the second day of the week with more announcements, which is consistent with the view that M&A plans often require a weekend to be finalised, as discussed in chapter 4. Occasionally, M&A deals that are being prepared during the weekends face some unforeseen difficulties that may force to postpone the announcement for Tuesdays. Using the same theory, that weekends are often used to work on finalising M&A deals, it can be consequently argued that Friday tends to be the less active working day, in terms of M&A announcements, as it may signal the beginning of a quieter period for executives to work on final agreements.

Finally, according to Branch et al. (2001) data, Wednesday and Thursday present similar levels of activity, in between a busier start of the week and a quieter Friday.229 Branch et al. (2001) data is consistent with the sample for 2000-2002 that was used in this thesis. Indeed, Monday accounted for a daily average of 29.4 announcements; Tuesday’s, 27; Wednesday’s, 24.9; Thursday’s, 24.6; and the weakest day, Friday, totalled only an average of 20.4 deals per day.

It is important to mention that effects from seasonality, holidays, trading days’ patterns, and other calendar issues, often appear interrelated. Whenever two or more factors occur simultaneously, the effects may result cumulative or dilutive. For example: in year 2000, the 4th of July holiday was celebrated on a Tuesday, which resulted in a four-day weekend, since this holiday is a non-trading day. In this case, much of the ‘end-and-beginning-of-the-month phenomenon’ was brought forward to the last day of June, a Friday, with 68 announcements, while the first day of July, a Saturday, also had an abnormal activity of 21

229 Vid. Branch et al. (2001) for a detailed data analysis of M&A announcements during working days for the period 1980-1999.

273 announcements.230 These figures are not only justified by the dilution of the ‘end-and-beginning-of-the-month phenomenon’, but also by the activity of the first two weekdays of July which have been also brought forward. In fact, Monday, 2nd July, which was a half day trading session, has had only fifteen announcements, while the following Tuesday, a holiday and non-trading day, has had merely two announcements. This M&A pattern of activity is also in line with several “day-of-the-week effects” which take place in the stock markets, namely the existence of the “holiday effect”, i.e. the occurrence of pre-holiday abnormal activity, even if this effect seems to be diminishing after 1987 (Keef & Roush, 2005); alongside with the so-called “weekend effect”, i.e. the tendency for abnormal returns to be concentrated around the turn of the week (Lakonishok & Smidt, 1988).

To handle seasonality, and some of the previously described calendar effects, an E_BoM variable, which signals the ‘end-and-beginning-of-the-month’ phenomenon, was added to the models using weekly and daily data. This dummy variable has value one on the days and weeks whenever the effect is visible, and zero value otherwise. A positive sign is expected for this variable.

Seasonal dummy variables were also included, in all models. These variables assume that the seasonal effect is unchanged year after year. Variables representing quarters, months, weeks, and weekdays, were added to the different models, exploring the possible existence seasonal patterns in data, such as the abnormal “Monday effect” on M&A activity and on the financial markets (e.g. Branch et al., 2001), the around the “turn of the year effect” (Lakonishok & Smidt, 1988), the “January effect” (Haugen & Jorion, 1996; Keim, 1983), or simply specific “ordinary-day-of-the-week” effects (vid. e.g. Berument & Kiymaz, 2001; Galai & Kedar-Levy, 2005; M. Gibbons & Hess, 1981). For example, in the models using monthly data, a Jan variable assumes value one if the month is January, and zero otherwise. To avoid perfect multicollinearity among the different subsets of seasonal dummy variables, which would make it

230 The number of announcements on 30 June 2000 was the highest on an event window of several months. The number of announcements on 1 July was also exceptional for a Saturday.

274 impossible to compute the regression solution, one dummy variable has been left for each subset. Accordingly, Quarter4 in quarterly data models, February in monthly data models, Week5 in the weekly data model, and Friday in the daily data model, have been left from the long list of variables.231 As a result, three variables representing quarters, eleven variables representing months, four variables representing weeks, and another four variables representing weekdays, were added to the long list of variables of the models using quarterly, monthly, weekly, or daily data, respectively.232 intervention variables: To control whether FASB’s new M&A accounting rules have had a significant impact on the M&A activity, two event dummy variables were created. These two variables-of-interest were added with the purpose of detecting any potential effects surrounding the interventions caused by FASB actions. An intervention takes place whenever there is some external influence at a particular time, which affects the dependent variable (Makridakis et al., 1998: 271). In the present research, these variables are also referred to as “event”, or “target”, variables.

As discussed earlier in this thesis, on 14 February 2001, FASB published a revised exposure draft, which contained the final proposals for a new M&A accounting. That document confirmed the tentative decision on a ban on pooling of interests, first announced on 6 December 2000, and introduced the replacement of purchased goodwill amortisation charges by impairment tests. The new proposals could result in an anticipation of M&A activity, with the purpose of avoiding the new accounting rules that would be enforced in the following summer. To capture this possible effect, a dummy variable has been

231 “Week5” variable has value one in the last week of a month, only if that month has five Thursdays. In some cases, Week5 variable includes the first day of the following month. Since this variable is not as consistent as the remaining seasonal dummies employed in the model using weekly data, it has been chosen to do not include it in the long list of variables. 232 Seasonal dummy variables in models using quarterly data: Quarter1, Quarter2, Quarter 3; monthly data: Jan, Mar, Apr, May, Jun, Jul, Aug, Sep, Oct, Nov, Dec; weekly data: Week1, Week2, Week3, Week4; and daily data: Mon, Tue, Wed, Thu.

275 prepared, Event_ED, with zero value before the 14 February 2001 and value one after this date.233 As for the preceding 6 December 2000 FASB’s announcement, it was merely tentative and therefore there was absolutely no need for companies to react, as they could continue to use the same business combinations accounting rules for an undetermined period.

Following the publication of the revised exposure draft, in June 2001 FASB issued two standards: SFAS 141, effective 1 July 2001, and SFAS 142, effective 16 December 2001. Although SFAS 142 has been made effective only in December, the document states that goodwill acquired in business combinations after June 30, 2001 shall not be amortised. Therefore, in practical terms, this means that both standards produce effects from 1 July 2001. To capture any possible effects resulting from the effectiveness of FASB’s standards, a dummy variable Event, with zero value before 1 July 2001 and value one after, was added to the long list of variables. lagged variables: Finally, MA_lag, Log MA_lag, and WITH_lag, are lagged variables by one period of dependent variables MA, Log MA, and WITH, which have been included to handle residuals’ autocorrelation.234 This resulted in a reduction of one observation in the total of observations used in the models.

7.6 Univariate descriptive statistics

In this section, some univariate descriptive statistics are shown, which summarise the datasets used, and also enabling to examine some individual variables separately. The ranges of values, central tendencies, or data distributions, are some examples of parameters examined in the following paragraphs.

233 This variable is not used in the models using quarterly and monthly data, since the announcement date is not placed next to the end, or the beginning, of a quarter or month.

234 These lagged variables are shown in the equations as MAt-1, Log MAt-1, and WITHt-1.

276

Table E.1, exhibited in Appendix E, reports univariate descriptive statistics for unadjusted and non-dummy variables for the whole period of study, covering the 2000-2002’s and the 1994-2008’s samples. More precisely, panels A to C cover the 2000-2002’s sample at different levels of aggregation, while Panel D reports univariate descriptive statistics for the larger period of 1994-2008.

Univariate descriptive statistics for variables employed in the models using monthly data are shown in Panel A. Apart from interest rates variables, Fed and Fed Av, all variables present low values of standardised skewness and standardised kurtosis. IP and GDP have progressed steadily during the period of study, and therefore present the smallest coefficients of variation, while Fed variables exhibit the highest values, due to numerous swift changes in Fed’s monetary policy. In terms of dependent variables, WITH presents a higher coefficient variation than MA, a situation that can be regarded as normal, due to the reduced number of M&A cancellations, when compared to the number of M&A announcements. The quartiles distribution also seems normal for all variables, although Fed variables present lower and upper quartiles very close to minimum and maximum values, respectively. A further analysis of the histogram and the density trace for this variable, not shown in this thesis, revealed the existence of a double top, i.e., a high concentration of observations around 2% and 6%, respectively.

The situation of the interest rates is indeed a particular one, as the Federal Reserve, the central bank of the USA, commonly referred to as the “Fed”, was very active, in terms of monetary policy, during 2001 and 2002. The US economy was in risk of recession and, in 2001, the Fed started aggressively to cut interest rates. If the average interest rates were above 6% in 2000, by the end of 2001 they were lower than 2%. In 2002, the interest rates would suffer further reductions, remaining at historically low levels.

Panel B exhibits univariate descriptive statistics for variables employed in the model using weekly data. The analysis is similar to the one made for Panel A.

277

In terms of coefficients of variation, it has increased for MA when compared with monthly data, while Fed continues to exhibit the highest coefficient.

However, some univariate descriptive statistics deserve particular attention. Apart for MA, the values of standardised kurtosis are outside the range of -2 to +2 for all variables. The range departure is also true for MA, but only in terms of standardised skewness. This situation suggests possible departures from normality, given that whenever standardised kurtosis and skewness value are outside the range there is a greater possibility of invalidation of the statistical tests in respect to standard deviation.

On the other hand, this situation can be regarded as normal, since weekly data is significantly more random than monthly data. Therefore, variables exhibiting signs of a possible departure from normality, as a result, for example, of asymmetric distributions, are to be expected, since extreme observations became more evident in weekly datasets. Furthermore, the outside values of standardised kurtosis and skewness analysis can be relativised, as the number of observations in the model using weekly data is sizeable, increasing the likelihood of a normal distribution of the dataset used.235

In conclusion, for the models using low levels of data aggregation in this thesis, indicators such as standardised skewness and standardised kurtosis are not meaningful, as they involve larger datasets. Nevertheless, they may signal the need for addressing statistical issues, regarding the subsequent models validation. The analysis of the daily dataset follows below.

Univariate descriptive statistics for variables employed in the model using daily data are presented in Panel C. This panel is divided in sections (1) and (2). Section (1) shows univariate descriptive statistics for the whole period, 2000-

235 Vide the central limit theorem, which provides an explanation for the prevalence of the normal probability distribution whenever is used a sufficiently large number of independent random variables, each with finite mean and variance, justifying the approximation of large samples to a normal distribution in controlled experiments (Rice, 1995).

278

-2002, while section (2) is centred on the analysis of the periods that preceded and followed the effectiveness date of the new M&A accounting rules.

In terms of univariate descriptive statistics for the whole period, and comparing with the values of the statistics previously examined, a further deterioration has been registered. The coefficient of variation for MA has increased, and almost all values of standardised kurtosis and standardised skewness are outside the range. An expected outcome, as if weekly data proved to be more random than monthly data, it would be then for daily data to be on the extreme side of randomness. Accordingly, statistics such as standardised kurtosis and standardised skewness become even less relevant for the model using daily data, and its consideration can be relieved, as justified before.236

The observation of univariate descriptive statistics for the period -391 to +391 days around 1 July 2001, shown in section (2), reveals better indicators for the period preceding the effectiveness of the new accounting rules. In 2001, uncertainty ruled over markets and economy in the USA, and turbulence increased with the September 11 terrorist attacks. In addition to volatility, ex post event figures for M&A activity, stock prices indexes, and interest rates, were significantly lower. Consequently, the coefficients of variation increased for all variables, post eventum. This increase was particularly noteworthy for interest rates, as result of the previously described Fed policy. A minimum and maximum value of interest rates of 3.68% and 7.03%, before 1 July 2001, compares with rates in between 1.15% and 3.89% afterwards.

Finally, Panel D shows univariate descriptive statistics for the period from 1994 to 2008. Overall, and despite covering a much larger period, the results of the analysis for quarterly data are similar to the ones for the remaining dataset aggregations. For example, the coefficients of variation exhibit similar figures. There are also quarterly variables presenting values outside the expected range of standardised skewness and standardised kurtosis, namely IP, and WITH variables, respectively, for which the analysis made before applies.

236 Ibidem.

279

7.7 Empirical tests and discussion of results

The statistical testing performed in this thesis has been made using the SAS 9.1 program, as previously mentioned, and using as well STATGRAPHICS Centurion XV, and STATISTICA 6.0.237 The purpose of using different statistical software packages is twofold. Firstly, it allows to extend the number of statistical tests that can possibly be made. Although comprehensive, these programs include only a limited set of statistics among the vast range of existing tests and procedures. They also tend to be specialised in particular subsets, to the detriment of others. Although the combination of the three programs provides a large set of statistical tests, some computations were also computed manually (e.g. Theil’s U statistic, from a holdout set). Secondly, the use of different programs, simultaneously, makes it possible to check and cross the outputs obtained using different programs, assuring the absence of computing mistakes.

7.7.1 Bivariate analysis

Following the previous summarisation and description of main data and variables used in this thesis, in this section a bivariate analysis is conducted, concerned with testing the relationships between pairs of variables, particularly between the dependent variables and the corresponding independent variables, in the context of the different models that have been developed. This analysis is intended to evaluate whether or not there exists an association, and the strength of such association, between specific sets of two variables. On the one hand, it is relevant to ensure that the explanatory variables are, in general, significantly related to the dependent variables, i.e. ensuring they have some relevant explicative power in the models. On the other, in the case of the

237 STATGRAPHICS Centurion XV, henceforth referred as Statgraphics, is a statistical software package copyright of StatPoint, Inc (2006). STATISTICA 6.0, is a statistical software package copyright of StatSoft, Inc. (2001).

280 independent variables, it is relevant to examine whether they exhibit high levels of association among each other, suggesting the inexistence of significant differences between variables, and therefore indicating a possible situation of redundancy in the use of a variable. In this case, it is also relevant to assess whether such high correlation is benign, in terms of model fitting.

Table F.1 presents Pearson and Spearman correlation coefficients for several variables taken from the long list. Correlation coefficients estimate the strength of the linear relationship between the variables. The correlation coefficient ranges between -1 and +1, representing -1 a perfect negative linear relationship, while, conversely, +1 indicates a perfect positive linear relationship. Coefficient values around zero indicate absence of linear relationship between variables. P- -values, which test the statistical significance of the estimated correlations, are exhibited in italic in Table F.1. P-values below 0.05 represent statistically significant non-zero correlations at the 95% confidence level. Unlike Pearson coefficients, Spearman correlation coefficients are measured from ranks of data values rather than from data values themselves. As a result, Pearson coefficients are more sensitive to outliers than Spearman coefficients.

A Pearson/Spearman correlation matrix of coefficient estimates for non-dummy variables using monthly data is shown in Panel A. Apart from coefficients for pairs of variables involving TD, all correlations are significant at the 95% confidence level.238 The only significant correlation for TD, at the 95% level, is with MA, if measured by Pearson correlations only. TD is also the variable with the lowest correlation coefficients. This may be justified by the specific nature of this variable, which has the characteristics of an adjustment variable, and consequently does not share the attributes of the remaining exogenous explanatory variables considered in the long list of variables.

Dependent variable MA presents higher correlation coefficients with the corresponding explanatory variables, than dependent variable WITH.

238 More precisely, except for TD, all pairs present statistically significant non-zero correlations at the 99% confidence level.

281

Explanatory variables related to stock markets indexes and capitalisation, SP500, SP500 Av, and MKTC, are the ones who present the highest correlations with the dependent variables, MA and WITH.

In terms of overall correlations between independent variables, the highest coefficient values belong to variables that share similar features. It is the case of the quasi-identical pairs of variables, Fed and Fed Av, and SP500 and SP500 Av, which are measured by final and average monthly values, and therefore share the same basis of construction. It is also the case of stock markets variables, which include stock prices index variables, SP500, and SP500 Av, and market capitalisation variable, MKTC, because they share the same nature. As mentioned earlier in this chapter, it is expected that stepwise regression will select the most appropriate set of variables from the long list of variables, by eliminating the most redundant variables. Otherwise, a manual elimination may be necessary to ensure the validity of the testing models.

Finally, as discussed earlier, time variables, Per, Per_2, and Per_3, are linearly related and therefore present high mutual Pearson correlation coefficients. Spearman correlations capture the linear dependence between variables in a different way. For the pairs of time variables, Spearman correlation coefficients assume +1 value, indicating, as expected, a perfect positive linear relationship between the variables.

A similar scenario is made visible in Panel B, where mutual correlations for non-dummy variables to be tested in the model using weekly data are shown. All pairs of variables present statistically significant non-zero correlations at the 99% confidence level, except for mutual correlations involving TD. The only significant correlation involving TD is with MA, also with a non-zero correlation statistically significant at the 99% confidence level, if measured by Pearson correlations, and still significant, but at the 95% confidence level, if measured by Spearman correlations. Exogenous explanatory variables, namely stock prices index and interest rates, continue to present high correlation coefficients, similar to time variables as well. As in Panel A, in terms of overall correlations, the

282 highest coefficient values belong to subsets of variables that share the same nature, or have identical basis of construction, such as the case of the variables constructed as linearly dependent.

Like panels A and B, Panel C shows correlation coefficients for variables employed in models developed to test hypothesis one, but using daily observations instead. Correlation coefficients for dummy variables are also shown, except for weekday’s dummy variables.

Every pair of non-dummy variables presents statistically significant non-zero correlations at the 99% confidence level, while dummy variables only present statistically significant correlations with dependent variable MA. The correlation coefficients, which were inferior in weekly variables when compared to monthly variables, exhibit even lower values, as expected. Nevertheless, exogenous explanatory variables and time variables continue to present correlation coefficients not discreditable, similarly to dummy variables as well. The correlation of dummy variable E_BoM and dependent variable MA constitutes a good example, as it presents a Pearson correlation coefficient similar to the ones presented by pairs of MA and non-dummy variables, namely, SP500, Fed, and period. Most importantly, the majority of the explanatory variables continue to be correlated with MA at the 99% confidence level.

Finally, Panel D shows correlations between variables constructed for a long run testing. Interestingly, during the period 1994-2008, the correlation between M&A activity and stock prices was more weak than in every model used to test the period from 2000-2002, including the model using daily data. It appears that a longer period of data aggregation - this is, quarterly versus monthly, weekly, or daily – may oversmooth volatility in data, weakening the likely pairwise behaviour of M&A activity and stock prices. Nevertheless, the MA and SP500 pair exhibits the second highest correlation value, behind the pair MA and Fed. Overall, the majority of the independent variables are correlated with MA and WITH variables, at least at the 95% confidence level.

283

The results obtained from the bivariate analysis corroborate, in general, the variables predictions and the theory that has been discussed previously. GDP variable is the only exception, as a positive correlation with MA would be expected. Nevertheless, it has been referred that M&A activity is procyclical, as generally it leads the business cycle (see e.g. Golbe & White, 1988; R. L. Nelson, 1959; Steiner, 1975; Weston et al., 1990). Consequently, the possible lag between M&A activity and GDP may have resulted in this outcome. It is feasible to admit this possibility, knowing that M&A waves and business cycles lengths are often long, and that a lag of several months may exist between these two different series. In any case, an in-depth examination of the reasons behind this possible contradiction is not relevant for the present research. In addition, following the backward elimination procedure that has been applied to the long list of variables, GDP variable has been selected for one model only. Consequently, the risk of model misspecification as a result of GDP inclusion is limited to a single model.

7.7.2 Multivariate analysis

The following multivariate analysis serves to corroborate the predictions made for explanatory variables. It also allows confirmation of the indications obtained from the bivariate analysis. This is made possible by the analysis of the coefficients exhibited by the variables which have been selected by the stepwise regression models. Nevertheless, the main purpose of the multivariate analysis is to examine whether event variables are selected for the models. In addition, a main concern is to verify if the data employed does not violate some key regression assumptions, namely, residuals normality, absence of serial correlation, and homoscedasticity. If these assumptions are not observed, then some parameters of the results, such as explicative power or confidence intervals, may be flawed, and in some extreme situations may even be seriously biased or misleading. Accordingly, a wide range of statistical tests was performed according to what is believed to be an appropriate diagnosis. Together, these tests also contributed for fitting the models, helping to choose

284 an appropriate functional form, and preventing some possible serious misspecification.

Due to the large amount of statistics that this section includes, the exhibition of the outputs had to be arranged following these criteria: tables resuming key statistics are mostly included in the body text, while full disclosure, using larger tables and additional illustrations, is placed in appendices.

In Appendix G the regression outputs for the models are exhibited, together with results from conformity tests. The outputs for equations (19), (20), (21), and (18) are shown in Table G.1, respectively on panels A, B, C, and D. All these models were constructed with the purpose of testing hypothesis one. Concurrently, Table G.2 exhibits the outputs for model equations (22) and (23), which have been created to test hypothesis two.239

Stepwise regression, with backward elimination, has been employed in all models, resulting in every variable left in the models to be significant at least at the 0.05 level. Most of the variables were created for specific models, resulting in quarterly, monthly, weekly and daily variables. Only five variables were considered in all models for a possible selection: SP500 and SP500 Av, Fed and Fed Av, and intervention variable Event. As discussed earlier, the relevant intervention variable Event_ED was not used in the models employing quarterly and monthly data due to unfeasibility.

Following stepwise backward elimination, the stock prices index variable, represented by SP500 and SP500_Av, was the most selected overall, corroborating existing literature, as mentioned recurrently in this thesis. More importantly, the intervention variables were not selected by any model. Consequently, at this stage the research hypotheses cannot be rejected in the null form. Nevertheless, it is very important to assess about the consistency of this critical finding. Accordingly, the remainder of this chapter is devoted to

239 Table G.3 resumes the regression analysis results for hypothesis one and two, providing a condensed examination of the variables selected for every model developed in this thesis.

285 checking the robustness of the models and the validity of this finding, by means of testing regression model assumptions and other statistical issues, development of sensitivity analysis tests, and testing of predictive power, in order to control whether under different circumstances and facing possible model violations, such intervention variables could be selected for the models, possibly preventing to assume the preliminary finding that the research hypothesis cannot be rejected in the null form.

As mentioned before, Panel A, in Table G.1, presents the results for the model constructed to test hypothesis one, using monthly data. From the twenty-four variables initially considered, only eight have been selected for the final model, as a consequence of the backward elimination of sixteen variables. Fed and Per_3 variables are significant at the 0.05 level, while the remaining selected variables are also significant at the 99% confidence level.

In terms of analysis of variance, F-ratio computed from ANOVA table is 68.49, with a p-value less than 0.05, therefore indicating a statistically significant relationship between the selected variables at the 95% confidence level.

In terms of goodness-of-fit for the model, the R-Squared statistic is 0.9546, which indicates that the final model explains 95.4% of the variability in MA. Nevertheless, the adjusted R-squared statistic is more suitable for comparing models with different numbers of independent variables. The adjusted R- -squared is a modification of the standard R2. It simply adjusts for the number of explanatory terms that a model includes. R-squared may result misleadingly, because when a model employs many variables, some unrelated explanatory variables’ variations may explain small portions of the variation of the dependent variable. Therefore, the adjusted R2 is often a more consistent indicator. In this model, the R-squared statistic adjusted for degrees of freedom is 94.07%. That means that the eight variables selected by the backward regression account for 94.07% of the explained variance in the model, which represents a significant percentage. Regarding the standard error of the estimate, it exhibits a standard deviation of the residuals of 28.05, while the

286 mean absolute error (MAE), which measures the average value of the residuals, is 19.33.

In Panel B, the regression outputs for the model using weekly data are shown. The stepwise backward regression procedure resulted in the elimination of seven variables from the initial sixteen. Nine variables have therefore been selected for the final model. Every variable that was selected is significant at the 95% confidence level, and only Fed and Week3 are not significant at the 99% level. The construction of ANOVA table provided an F-ratio value of 49.60, with a p- -value inferior to 0.05, which indicates the existence of a statistically significant relationship between the variables at the 95% confidence level.

The adjusted R-squared statistic indicates that the fitted model explains 73.95% of the variability in MA. The standard deviation of the residuals is 17.32, while the mean absolute error of the residuals is 13.40. The comparison of these statistics with the ones from the model using monthly data, allows one to observe a decrease on the predictive power of the model, as adjusted R-squared value has decreased. On the other hand, a reduction was recorded on the average value of the residuals and on the standard deviation of the residuals.

Panel C exhibits the statistical outputs for the model using daily data. From the initial fifteen variables, only four were removed: Fed, Per_3, Event, and Event_ED. All weekdays and remaining endogenous explanatory dummy variables were selected, being every variable significant at the 95% confidence level. While the selection of Per, Per_2, and Per_3 variables means this is a third-order, or cubic, polynomial regression model, the selection of weekly dummy variables and other endogenous variables confirm the existence of significant M&A patterns of behaviour, such as the case of the “Monday effect” and the “end-and-beginning-of-the-month phenomenon” (e.g. Branch et al., 2001; Thomson Financial, 2006). Like in previous models, SP500 keeps exhibiting a high level of significance. The lagged dependent variable, MA_lag, was also selected. The F-ratio from analysis of variance is 95.34. Since its p-

287

-value is less than 0.05, there is a possible statistically significant relationship between the variables at the 95% confidence level.

The variability of MA is explained in 57.09% by the final model. Although lower than the remaining models, it continues to constitute a significant percentage of explanation of the variability in MA. Comparing with the model using weekly data, alongside with the decrease on the adjusted R-squared statistic, a further reduction on the standard deviation of the residuals, from 17.32 to 6.69, and on the MAE of the residuals, from 13.40 to 5.14, has also been observed.

Finally, regarding testing hypothesis one in the long run, Panel D shows the regression output for the model using quarterly data. The stepwise backward regression procedure resulted in the elimination of nine variables from the initial thirteen.240 Every variable selected is significant at the 99% confidence level. The F-ratio ANOVA value is 66.13, with a p-value inferior to 0.05, which indicates the existence of a statistically significant relationship between the variables at the 95% confidence level. In this model, the R-squared statistic adjusted for degrees of freedom is 81.78%. The standard deviation of the residuals is a mere 0.0335, while the MAE exhibits also a residual value of 0.0262.

240 The use of stepwise regression resulted in the selection of two variables, SP500 and MKTC, which may raise concerns as they are both market variables. Nevertheless, the metrics used in these variables are different, as SP500 considers the closing market prices at the end of a quarter, while MKTC weights the total market value of a whole quarter. Furthermore, as to be demonstrated throughout this chapter, from a statistical viewpoint, the model fitting seems appropriate and valid. Therefore, the outcome of the stepwise regression was accepted. Nevertheless, in order to avoid any misjudgement, as a result of a possible biasing presence of these two variables together in the final model, some alternative regressions were run. In terms of results, had SP500, or MKTC, been manually eliminated, and the adjusted r-squared would be reduced to 78.22% or 75.40%, or, respectively. Regarding the regressions analyses shown in this thesis, the results of such alternative testing did not indicate the need for any significant adjustments.

288

As for hypothesis two, the results are shown in Table G.2. Panel A shows the estimation and tests results for equation (23), which uses monthly data, and WITH as dependent variable. From the initial twenty-five monthly variables, nine have been selected for the final model, to the detriment of the other sixteen, which have been eliminated following the backward elimination procedure. TD categorical variable and Oct dummy variable are significant at the 95% level, while the remaining selected variables are also significant at the 99% level. The ANOVA F-ratio is 20.80, and the p-value is less than 0.05, suggesting a statistically significant relationship between the variables at the 95% confidence level. The adjusted R-squared statistic indicates that the model as fitted explains 83.976% of the variability in the number of withdrawn M&A deals. In respect to residuals, the standard deviation is 3.6517, and the MAE is 2.6611.

Finally, Panel B shows the regression output for the model testing the long run and using quarterly data. Nine out from fourteen variables initially considered were eliminated. Therefore, five variables have been selected for the final model, namely SP500_Av, Fed, MKTC, IP, and TD. Apart TD, significant at the 95% level, all other variables are significant at the 99% confidence level.241 F-ratio ANOVA value is 65.03, with a p-value inferior to 0.05, suggesting a statistically significant relationship between the variables at the 95% confidence level.

The variability of WITH is explained in 84.66% by the fitted model, a significant percentage. The standard deviation of the residuals is 13.75, and the MAE is 10.42, values several times superior to the three models examined immediately before. Nevertheless, they compare well with the high values of models testing hypothesis one and using monthly and weekly data. The existence of possible misspecifications and statistical violations is therefore to be analysed in the following sections of this chapter.

241 Ibidem, with the difference that in this case SP500_Av was chosen instead of SP500. Had SP500_Av, or MKTC, been manually eliminated, and the adjusted r-squared would be reduced to 75.44%, or 82.11%, respectively

289

autocorrelation: Regression with time series involves several assumptions that need to be checked (see e.g. Dougherty, 2002; Gujarati, 1995; Lomax, 2007; Makridakis et al., 1998; D. C. Montgomery et al., 2001; Stevens, 2002). The existence of autocorrelation is one of the possible serious violations of the statistical assumptions underlying the regression analysis. The examination of a possible missing independence in the residuals is therefore critical.

It is a fact that the assumption of uncorrelated, or independent, errors for time series data is not often appropriate, since it is usual that errors in time series data exhibit serial correlation (see e.g. D. C. Montgomery et al., 2001). Furthermore, is no less true that serial correlation in the residuals indicates that the regression model is possibly misspecified or, at least, that its fitting could be improved.

The analysis of the autocorrelations and partial autocorrelations is as important in an early stage of the models construction, as it is afterwards. In the present research, were analysed estimated autocorrelations and partial estimated autocorrelations using Statgraphics.242 The examination of autocorrelations in residuals is critical to perceive whether the time series may not be completely random. In time series, random numbers are often referred as noise. If residuals are not random, this may indicate there are interactions the model was not able to capture. In estimated autocorrelations, the lag k autocorrelation coefficient measures correlations between values of residuals at time t and time t-k.

Partial autocorrelations can be used to measure the degree of relationship between lagged variables, Yt and Yt-k, when the effect of other time lags is removed, helping to determine the order of autoregressive model needed to fit the data. It has been previously mentioned that the long list of variables includes, for every model, a lagged dependent variable, by one period. In the

242 The following conclusions and definitions are mostly taken from reports and tables produced by Statgraphics (StatPoint Inc., 2005, 2006).

290 tables produced by Statgraphics, the lag k partial autocorrelation coefficient measures correlations between values of residuals at time t and time t+k, having previously accounted for the correlations at all lower lags.

Exhibited in Appendix H are the estimated autocorrelations and the estimated partial autocorrelations tables for the models used in this thesis. Figures with autocorrelations correlograms are also included, enabling an easier visualisation.

Table H.1 presents the autocorrelation tables for the models used to test hypothesis one. The figures on these tables are also exhibited as correlograms in Fig. H.1. For the model using monthly data, exhibited in Panel A, the estimated autocorrelations coefficients are contained within the 95% probability limits for the eleven lags defined by default in Statgraphics, accordingly to the sample size, meaning that none of the autocorrelations coefficients is statistically significant at the 0.95 level. This outcome indicates that the time series may well be completely random, being equivalent to a white noise series.

As for the model using weekly data, which autocorrelations are shown in Panel B of Table H.1, and in Fig. H.1, the findings are identical to the model using daily data. All twenty-four coefficients are contained in between the 95% probability limits. Since the autocorrelations are not statistically significant at the 0.95 level, the time series is therefore likely to be completely random. Similarly, as shown in Panel D, all the nineteen coefficients of the model using quarterly data are within the 95% threshold.

In terms of estimated autocorrelations for residuals in the model using daily data, three of the twenty-four autocorrelations coefficients are statistically significant at the 95% confidence level, suggesting that the time series may not be entirely random. Nevertheless, when probability limits are eased to 99%, none of the autocorrelations coefficients are statistically significant, indicating, in opposition, that the time series may well be completely random, but only at a lower confidence level.

291

As it is possible to verify in Table H.2 and in Fig. H.2, the autocorrelations figures for residuals in the models used to test hypothesis two, reveal that all coefficients are contained in between the 95% probability limits. Since the autocorrelations are not statistically significant at the 0.95 level, the time series may well be therefore equivalent to white noise series.

In terms of partial autocorrelations, like in the case of the estimated autocorrelations, most of them are contained in the 95% probability limits, as shown in Table H.3 and Table H.4. The exceptions are one partial autocorrelation for each of the models using weekly and quarterly data (hypothesis one), and two partial autocorrelations for the model using daily data. When the probability limits are eased to 99%, none of the autocorrelations coefficients are statistically significant, indicating that the time series may well be entirely random.

The global analysis of the autocorrelations and partial autocorrelations at the several lags provides no indications of serious autocorrelation in the residuals, except perhaps for the model using daily data. In order to investigate further this possibility, and also to corroborate the absence of serious autocorrelation in the models, other statistics were performed as follows.

Durbin-Watson (DW) is a classic statistic, used to detect the presence of autocorrelation in the residuals. It also tests for a possible misspecification of the models, as a result of an inadequate fitting. It tests residuals in order to capture any significant correlations in data, based on the order in which residuals arise in the series that have been utilised. More precisely, the Durbin- -Watson statistic tests for first-order serial correlation in the residuals of a time series regression. Its value has the range from zero to four. A value less than two suggests a positive autocorrelation, while a value greater than two indicates a negative autocorrelation. A commonly rule of thumb given for Durbin-Watson

292 statistic is that a value of approximately two indicates that there is no serial correlation.243

From the observation of Table G.1 and Table G.2, shown in Appendix G, it is possible to observe that, for every model, Durbin-Watson d-statistic present values close to the statistic reference value, and p-values greater than 0.05 as well. For the model employing quarterly data used to test hypothesis one, d- -statistic value is 2.005, with a p-value is 0.31. Since the p-value is greater than 0.05, there is no indication of serial autocorrelation in the residuals at the 95% confidence level. The models employing monthly data and daily data exhibit a similar condition. Finally, although presenting a d-statistic more distant from the reference value, 1.808 versus two, the p-value for the model using weekly data of 0.11729 is greater than 0.05, therefore suggesting the inexistence of significant serial autocorrelation in the residuals at the 95% confidence level as well. For the models developed to test hypothesis two, as shown in Table G.2, the same condition: Durbin-Watson d-statistic is also more distant from the reference value, but has p-values greater than 0.05. Since every p-value is comfortably greater than 0.05, there is no indication of the existence of serial autocorrelation in the residuals at the 95% confidence level for all regressions shown in this thesis.

The interpretation of Durbin-Watson statistic can be more or less conservative. For instance, Ott (1992) argues that if the statistic value ranges within 1.5 and 2.5 there is no serious problem of autocorrelation. Nevertheless, a more accurate interpretation lies in the examination of lower and upper critical values of Durbin-Watson d-statistic as exhibited in Appendix I (see e.g. Brooks, 2005; Dougherty, 2002; Makridakis et al., 1998). Fig. I.1 illustrates the five regions of the Durbin-Watson d-statistic, which can be computed using DWL and DWU 244 values. If d-statistic is lower than DWL, or higher than 4-DWL, it indicates the existence of positive, or negative, significant autocorrelation, respectively. If d-

243 Information about DW taken from Brooks (2005), Dougherty (2002), Makridakis et al. (1998). 244 DWL and DWU for Durbin-Watson d-statistic lower and upper critical values, respectively.

293

-statistic lies in between DWU and 4-DWU there is no indication of autocorrelation. Finally, if it is between DWL and DWU, or between 4-DWU and

4-DWL, the test is inconclusive. Table I.1, also includes the critical values used in this thesis for the examination of the Durbin-Watson statistic.

Table 7.1 First lag autocorrelations and Durbin-Watson statistic values Hypothesis One Hypothesis Two Quarterly Monthly Weekly Daily Quarterly Monthly

First lag autocorr. -0.022356* -0.05281* 0.09207* -0.00482* 0.009070* -0.18472*

Durbin’s h -0.2086* - - -0.22345* - -

Durbin-Watson d 2.00502*‡ 2.07124*† 1.80846*† 2.00652*‡ 1.9579* 2.28273*† * No significant autocorrelation at the 5% significance level † Inconclusive test using critical values for Durbin-Watson (see Table I.1) ‡ Inappropriate test since the model employs a lagged dependent variable

Table 7.1 presents first lag autocorrelations and Durbin-Watson statistics for all models. As exhibited in Appendix H, apart from one model using weekly data, and another one using quarterly data, all first lag autocorrelations are negative. Table 7.1 also shows the Durbin-Watson d-statistic values as reported previously in Appendix G. No significant autocorrelations at the 5% significance level has been found. The significance levels were also measured using critical DW values, as shown in Appendix I, in accordance to the interpretation of the five regions of the Durbin-Watson d-statistic.245 As exhibited in Table 7.1, the interpretation comes inconclusive for the models using monthly and weekly data. For the models using quarterly and daily data, the analysis of DW critical values corroborates the previous indication of absence of any significant autocorrelation in the residuals.

245 A measurement and interpretation example for the model using weekly data follows. Given that there are nine explanatory variables and 155 observations used, we have k=10 and n=155, and critical values DWL=1.616 and DWU=1.861. Since the test statistic, 1.808, is greater than the lower critical value, but lower than the upper critical value, it is not therefore possible to conclude if there is any lack of independence in the errors.

294

The long list of explanatory variables includes lagged dependent variables. One of the main purposes of such inclusion is to correct potential AR errors. Nevertheless, whenever the regression incorporates lagged dependent variables, the Durbin-Watson d-statistic is inappropriate to test for autocorrelated residuals, because it tends to be biased towards two, its reference value. That is the case of the models employed to test hypothesis one, and using quarterly or daily data, where the stepwise backward elimination resulted in the selection of the lagged dependent variable.

In case a model uses a lagged dependent variable by one period as an explanatory variable, Durbin's h test must be considered (see e.g. Dougherty, 2002).246 The appropriate critical value for Durbin’s h at 5 % level of significance is ±1.96. Therefore, for a test of the null hypothesis of no autocorrelation against the two-sided alternative of autocorrelated errors, the decision rule is if

-1.96 < h < 1.96 then do not reject the null hypothesis. Since the computed Durbin’s h values of -0.2 are within the critical values of a standard normal distribution, the null hypothesis of no evidence for autocorrelation in the residuals cannot be rejected as well.

Finally, additional specific tests for randomness of residuals were performed. These kinds of tests are used to examine whether residuals consist in a random sequence of numbers. Using Statgraphics, three main tests were run:247 i) the first test counts for the number of times that the residuals sequence was above or below the median. In the case of the model using monthly data, with the purpose of testing hypothesis one, the number of runs was twenty, versus an expected number of eighteen. The large sample test statistic Z for this regression model is 0.5224, with a p-value of 0.601338; ii) the second test counts the number of times the residuals sequence rises or fall. The number of runs up

246 The Breusch-Godfrey test, which covers autocorrelations of higher orders, could be used alternatively. This test has the advantage of remaining appropriate even when regressors may include lags of the dependent variable, However, this statistic it is not reliable for relatively small samples. 247 The tables in Appendix G exhibit the results for Box-Pierce test only.

295 and down was twenty-three, which matches with the expected number of runs up and down. Test statistic Z value is 0.5224, with a p-value of 0.536876; and finally, iii) Box-Pierce test which is, by default in Statgraphics, based on the sum of squares of the first twenty-four autocorrelation coefficients.248 Nevertheless, in this case, the small sample size makes it possible to perform the test based on the first eleven autocorrelations only. The p-value for this test is 0.630565. Since every p-value for the three tests performed for the equation (19) is greater than 0.05, the hypothesis that the series is random at the 95%, or higher confidence level, cannot be rejected.

As for the model examined above, the same three tests of randomness of residuals were performed for the model using weekly data. For equation (20), the runs above and below median test has a statistic Z value of 0.4042; runs up and down test has a statistic Z value of 1.2455; and Box-Pierce test exhibits the value of 23.701. All the p-values are greater than 0.05: 0.686037, 0.212927, and 0.478771, respectively. Consequently, the hypothesis that the series of residuals is random at the 95% confidence level cannot also be rejected. Similar results were also obtained for the model using quarterly data, with p-values of 0.691054, 0.27234, and 0.187742.

In the case of the model using daily data, two of the performed tests have p- -values greater than 0.05. It is the case of runs above and below median test, and runs up and down test. However, Box-Pierce test presents a p-value which is only greater than 0.01. Consequently, the hypothesis that the series is random at the 95% confidence level can be rejected. According to Statgraphics (StatPoint Inc., 2005, 2006), since the three tests are sensitive to different types of departures from random behaviour, failure to pass any test suggests that the time series may not be entirely random. Nevertheless, since the Box-Pierce test is still greater than 0.01, the hypothesis that the series is random cannot be rejected at the 99% confidence level.

248 Tests descriptions based on Statgraphics’s tests reports and user’s guide (StatPoint Inc., 2005, 2006).

296

Regarding the models used to test hypothesis two, the same three tests for randomness of residuals present p-values greater than 0.05. Therefore, it cannot be rejected the hypothesis that the series of residuals is random at the 95%, or higher, confidence level. normality: Another statistical assumption of the regression model is that the error term is normally distributed. Sometimes the error distribution is skewed, or may suffer from abnormal kurtosis, as a result of the presence of extreme observations. The regression parameter estimation is based on the minimisation of squared errors, and a few large outliers are enough to influence such estimates. In case the error distribution is significantly abnormal, the confidence intervals may be too much wide, or too narrow, and the significance tests may result inappropriate (see e.g. Makridakis et al., 1998).

A set of figures is shown in the Appendix J, with the purpose to allow the graphic visualisation of the residuals’ distribution. One of the immediate ways to assess whether errors are normally distributed, is to check if the residuals are scattered in a horizontal band with no values too far from such band, and to check whether no visible patterns exist, such as curvatures or increasing spreads (see e.g. Makridakis et al., 1998). The plots of residuals, in Fig. J.1 and Fig. J.2, exhibit no such patterns, despite some outliers that are made visible, particularly in the model using daily data.

The analysis of histograms of residuals proves to be very helpful for an improved insight concerning the residuals normality. Fig. J.3 and Fig. J.4 plot histograms of residuals, with a normal distribution superimposed, for models used to test hypothesis one and two, respectively. The histograms exhibit the number of residuals obtained within each of the intervals marked on the horizontal axis, while the normal curve shows how many observations would be obtained on average for a normal distribution (Makridakis et al., 1998: 263). According to the histograms shown in Appendix J, the normality assumption seems to be ensured.

297

Finally, Fig. J.5 and Fig. J.6 exhibit normal probability plots for residuals. Its visualisation is perhaps the best procedure, whenever using visual observation, to evaluate whether errors follow a normal distribution pattern. These figures plot the quantiles of error distribution versus the quantiles of a normal distribution having the same mean and variance. If the distribution is supposed to be normal, then the points on these plots should fall close to the 45º diagonal line. Deviations from that line suggest departures from normality. An s-shaped pattern of deviations from the diagonal indicates that the residuals have excessive kurtosis, while a bow-shaped pattern of deviations from the diagonal line suggests that the residuals have excessive skewness. In terms of analysis of the plots shown in Fig. J.5 and Fig. J.6, the graphical observation corroborates the previous indications of a normal distribution of errors for every model used in this thesis.

Besides the graphical analysis, a wide range of statistical tests was performed to test normality under the null hypothesis that the residuals are normally distributed. Some of the key statistics displayed in those tables are resumed below in Table 7.2.

298

Table 7.2 Key tests for residuals’ normality Hypothesis One Hypothesis Two Quarterly Monthly Weekly Daily Quarterly Monthly

Chi-Squared 15.2375 4.77137 33.0323 41.3188 14.5593 16.6575 0.578382 0.97988 0.16115 0.85599 0.62719 0.21544

Shapiro-Wilk W 0.97507 0.95536 0.96908 0.98240 0.98238 0.94590 0.483264 0.21262 0.03755 0.07773 0.77596 0.10997

Kolmogorov-Smirnov 0.08985 0.08167 0.05955 0.02704 0.05177 0.15148 0.72748 0.97369 0.64161 0.61745 0.99741 0.40288

Kuiper V 0.14359 0.15212 0.09618 0.04792 0.09383 0.22084 ≥ 0.10 ≥ 0.10 ≥ 0.10 ≥ 0.10 ≥ 0.10 ≥ 0.10

P-Values in italic.

The tests exhibited in Table 7.2 are commonly used to find whether residuals can be adequately modelled by a normal distribution. Given that small departures from normality may not affect the validity of analysis of the tests, it is therefore important to complement the examination of plots and key statistics with a wider set of tests for a final assessment of normality. Therefore, additional tests were also performed, as exhibited in Appendix G tables.

According to Statgraphics’s tests reports and user’s guide (StatPoint Inc., 2005, 2006): i) chi-squared test divides the range of residuals into sixteen equally probable classes and compares the number of observations in each class to the number expected based on the fitted distribution; ii) Shapiro-Wilk W test is based upon comparing the quantiles of the fitted normal distribution to the quantiles of the data; iii) standardised skewness test examines for lack of symmetry in the data; and finally, iv) standardised kurtosis test examines the distributional shape, which may be either flatter or more peaked than a normal distribution. Goodness-of-fit tests for residuals were also performed: v) Kolmogorov-Smirnov (K-S) test computes the maximum distance between the cumulative distribution of residuals and the cumulative distribution function

299

(CDF) of the fitted normal distribution;249 while vi) Kuiper V, Cramer-Von Mises W2, Watson U2, and Anderson-Darling A2 tests and modified forms, compare the distribution function to the fitted CDF in different ways.

The power of the statistical testing increases as the sample size becomes larger, making it increasingly easier to detect smaller departures from normality.250 Nevertheless, the power of the tests can also vary according to different sample size categories. For example, according to SAS Institute, Inc. (2004), for a sample size larger than 2000 observations, the Kolmogorov-Smirnov test is more suitable, while if the sample size is less than 2000, Shapiro-Wilk test should be used. On the other hand, there are researchers who argue that the Shapiro-Wilk test was conceived for sample sizes up to fifty observations, being other tests, such as the Anderson-Darling test, more adequate for testing samples sized from 51 to 1999 observations. Therefore, in the present research, Shapiro-Wilk test seems to be the most appropriate one for the models using monthly data, while other tests, such as the Anderson-Darling test, seem to be more adequate for the remaining models. Although the Kolmogorov-Smirnov test does not seem particularly appropriate for any of the models, it has been decided to keep it in the analysis, since it is a classical test to examine normality. It is also important to bear in mind that all these sample size considerations are merely rule of thumb. Accordingly, all tests ran are shown in this thesis to be considered at the discretion of the reader

Regarding the tests results for equations (18) and (19), all p-values obtained for chi-squared, Shapiro-Wilk W, standardised skewness, and standardised kurtosis, are greater than 0.05. Therefore, the hypothesis that residuals come from a normal distribution cannot be rejected at the 95% confidence level. The p-values of K-S test, modified Kolmogorov-Smirnov D (K-S D), Kuiper V and modified Kuiper V, Cramer-Von Mises W2 and modified Cramer-Von Mises W2, Watson U2 and modified Watson U2, and Anderson-Darling A2 and modified Anderson- -Darling A2 tests, are also all greater than 0.05, which, consequently,

249 For example, the maximum distance is 0.0816716 for equation (19). 250 Power of the test means ability to reject the null hypothesis of normality.

300 corroborates the previous indication of not rejecting the hypothesis that residuals come from a normal distribution with 95% confidence. This finding also corroborates the graphical observation made before.

As for the model using weekly data, except for Shapiro-Wilk W test, all tests performed to examine the normality of residuals and the goodness-of-fit tests for residuals: chi-squared test, standardised skewness, standardised kurtosis, K-S and modified K-S D tests; Kuiper V, Cramer-Von Mises W2, Watson U2, Anderson-Darling A2 tests, and modified V, W2, U2, and A2 versions, respectively; present p-values greater than 0.05. Since the p-value for Shapiro- -Wilk W test is only greater than 0.01, the hypothesis that residuals comes from a normal distribution can be rejected at the 95% confidence level, but cannot be rejected at the 99% level.

The lower p-value obtained for Shapiro-Wilk W statistic can be observed as normal. As mentioned before, whenever weekly or daily data is used, departures from normality become a normal situation. For example, Marais (1984) found that weekly residuals from CAR may deviate from normality with respect to skewness and kurtosis. In fact, a few outliers are often enough to deviate a series from normality. Moreover, since the sample size is greater than fifty observations, Shapiro-Wilk is not regarded as the most appropriate test. Accordingly, it seems to be arguable to admit that the residuals are normally distributed.

With regards to the model using daily data, in terms of goodness-of-fit tests for residuals, the hypothesis that residuals come from a normal distribution with 95% confidence cannot be rejected, given that the smallest p-value amongst the specific tests performed, such as the previously referred Kuiper V and K-S tests, is greater than 0.05.251 This conclusion is corroborated by other tests for normality for residuals, such as chi-squared and Shapiro-Wilk tests, which have p-values higher than 0.05 as well. Nevertheless, skewness and kurtosis Z-score’s

251 Apart Chi-Squared Test, every p-value is based on general tables and may be very conservative (StatPoint Inc., 2005).

301 exhibit very low p-values. Accordingly, it could be rejected the idea that residuals comes from a normal distribution with 95% confidence. Nevertheless, as mentioned before, the low p-values presented by standardised skewness and kurtosis match with the situation portrayed by authors such as Marais (1984).252 It is also important to point out that the regression results and tests presented in Appendix G include outliers, since its removal has been made at a later stage only. They will be examined in depth in the next section of this thesis.

Finally, in what concerns to the models used to test hypothesis two, apart modified K-S D test in the equation employing monthly data, all tests performed for the normality and goodness-of-fit tests for residuals, resulted in p- -values greater than 0.05. When generic p-values are used, instead of specific p- -values constructed specially for fitting the selected distribution, all p-values are greater than or equal to 0.10. Furthermore, Kolmogorov-Smirnov does not seem to be appropriate for this model, since it is more adequate for models employing more than 2000 observations. Accordingly, it seems that the hypothesis that residuals comes from a normal distribution cannot be rejected at the 95% confidence level, for both models using quarterly and monthly data. heteroscedasticity: Homoscedasticity, often referred to as homogeneity of variance, means that the error term has constant variance across observations. It is another assumption of the regression analysis. In opposition, heteroscedasticity is a violation of this assumption. When such violation occurs, it may affect the variance and the standard error of parameter estimates, the interval estimation, the hypothesis testing, Pearson coefficients, among other.

Heteroscedasticity often occurs when there is a large difference between sizes of observations. In the early stage of development, the model using daily data

252 As discussed earlier, departures from normality are common whenever weekly or daily data is used. Additionally, outliers can be kept in the models with no consequences in respect to the models validation. Therefore, the lower p-values for skewness and kurtosis do not seem to constitute a problem.

302 included weekends. Since the number of announcements at weekend days is clearly outnumbered by the activity occurred during business days, weekends had to be removed from the final sample.

The Test of First and Second Moment Specification was run, following White’s (1980) procedure. Its tabled values, exhibited in Table 7.3, were computed using SAS. The specification test assumes the null hypothesis that the error terms have homogeneous variance.

Table 7.3 Tests for heteroscedasticity Model Data DF Chi-Square Pr > Chi-Sq. Quarterly 14 13.35 0.4989 Monthly 34 25.95 0.8372 Hypothesis One Weekly 46 41.54 0.6593 Daily 60 73.28 0.1165

Quarterly 20 26.44 0.1518 Hypothesis Two Monthly 38 31.23 0.7737

As shown in Table 7.3, the lowest p-value of the Chi-Square test for the regression models is 0.1165. Therefore, the null hypothesis of a homoscedastic error variance is supported, as it has been found that residuals are not heteroscedastic at the 0.05 confidence level.

7.7.3 Outliers and influential points

As examined in the previous sections, overall, the models meet the basic regression assumptions. However, the impossibility of obtaining confirmations at the 95% confidence level in every occasion, suggest the existence of possible relevant outliers. The presence of outliers and high influential points increases the likelihood of violations of statistical assumptions, a situation that may lead

303 to biased or misleading results. Since no treatment has been given to outliers and influential points, like elimination of extreme observations, it is therefore recommendable to identify abnormal observations in order to evaluate any significant effects that may come from its presence in the models. outliers: By definition, an outlier is an individual data point which is inconsistent or very different from the remaining observations (Iglewicz & Hoaglin, 1993; Stevens, 2002), as it does not “fit the trend set by the balance of the data” (Myers, 1990: 221). It therefore consists in observations with large residuals (Makridakis et al., 1998).

The presence of outliers is a reason of particular concern, since many statistical measures, such as the mean and the variance, are sensitive to abnormal observations. Its presence in data can also contribute to the violation of statistical assumptions. Moreover, outliers can influence significantly the regression results: from R-squared values; to increases in the standard errors of the regression coefficients. It is therefore important to identify outliers, as it is important to evaluate its possible impacts on the regression results. If such impacts are significant, it may be needed to give residuals’ an adequate treatment, in order to ensure the robustness of the results.

Outliers may occur because of an error in data recording or an entry error, or simply because the subjects are different from the rest (Stevens, 2002, 2007).253 In the first case, raw data points should be checked, and corrected if possible, or deleted otherwise. If they do not result from an error in data, the outlier can be then considered for deletion but only in case of extreme observations, since there is also the possibility of modelling outliers separately. Robust regression techniques can also be used to reduce the impact of abnormal observations, avoiding its deletion.

253 Rare events and novel phenomena, inadequate distribution assumptions, or dynamic and unknown data structures, are other examples of possible sources of outliers, mentioned by other authors such as Iglewicz & Hoaglin (1993).

304

There are several techniques used to detect outliers, being the analysis of studentised residuals one of the commonest. Studentised residuals can be computed in different ways, one of which is to measure how many standard deviations each observed value of the response variable deviates from a model fitted using all data, except that observed observation (StatPoint Inc., 2006). This procedure results in an externally studentised residual, which is commonly labelled as R-student.

According to Myers (1990), an informal rule of thumb is that a R-student that exceeds +2 or -2 could be a potential outlier. Nevertheless, other authors argue that cut-off points of ±2.5, ±3, and even ±4, can be also acceptable in some conditions, although higher cut-off levels require more careful attention. Observations with studentised residuals greater than three can be considered for removal, or for separate handling.

Appendix K’ Fig. K.1 and Fig. K.2 show Box-and-Whisker plot of residuals for the models developed to test hypotheses one and two, respectively. In the Box- -and-whisker plot, the central box represents the values from the lower to the upper quartile, i.e., 25 to 75 percentile. The vertical middle line dividing the lower and upper quartiles represents the median. The horizontal line extends from the minimum to the maximum value, excluding any possible extreme points. These extreme points are shown as separate points, outside the boxplot, and can be classified as outside or far out values, i.e., mild or extreme outliers, respectively. An extreme point can be considered as a mild outlier if a value is smaller than the lower quartile minus 1.5 times the interquartile range, or larger than the upper quartile plus 1.5 times the interquartile range. An extreme outlier will be considered when a value is smaller than the lower quartile minus 3 times the interquartile range, or when a value is larger than the upper quartile plus 3 times the interquartile range.

The residuals’ quartiles distribution seems normal for all models, even if skewed in the case of the models testing hypothesis two, or using quarterly data (i.e.

305 three models). There are three plots exhibiting extreme points outside the boxplot: in the case of the models using quarterly data and in the model using daily data. These extreme points may be possible outliers. However, since they are smaller than the lower quartile minus 1.5 times the interquartile range, and larger than the upper quartile plus 1.5 times the interquartile range, they can be considered as possible “mild” outliers only.

Table K.1 and Table K.2 list the observations with studentised residuals greater than two in absolute value for all models, except for the model using daily data, for which only R-student greater than three are exhibited. In terms of hypothesis one’ models, the one using quarterly data has one studentised residual greater than ±3. The model using monthly data has two studentised residuals greater than ±2 and one greater than ±3, while the one using weekly data has nine studentised residuals greater than ±2, and two studentised residuals greater than ±3. The model using daily data includes thirty-four studentised residuals greater than ±2, six studentised residuals greater than ±3, and four greater than ±4, therefore raising some concerns.

Concerning the models used to test hypothesis two, the one using quarterly data has two studentised residuals greater than ±2 and one greater than ±3. Finally, the model using monthly data has two studentised residuals greater than ±2, but none greater than ±3. Therefore, in the case of these models, outliers do not seem to constitute a reason for concern.

Following the initial diagnosis, made possible by the use Box-and-Whisker’s plots, and the later examination of studentised residuals’, only the model used to test hypothesis and using daily data appears to constitute a reason for concern, since it includes several outliers outside the range, with many R- -student values greater than ±3.

Alongside with the R-student analysis, there are other tests used to identify outliers. The Rosner’s test, the Dixon’s test, and the Grubbs’ test, are examples of tests based on hypotheses testing that are commonly used for normal

306 distributions. Nevertheless, not all the tests may be adequate for the present research. It is the case of Dixon’s test, which is used for detecting small numbers of outliers (R. D. Gibbons, 1994) in small sets of data, usually comprising less than twenty-five observations. The Grubbs’ test is one of the most used for outliers and has therefore been employed in the present research.

Known as the maximum normalised residual test, the Grubbs’ test assumes that data can be reasonably approximated by a normal distribution. Since this condition is observed in the models of the present research, Grubbs’ test can be therefore applied. This test detects one outlier at a time. In case of multiple outliers, it becomes necessary to delete the single outlier detected and to run again the Grubbs’ test. This procedure is to be repeated until no outliers are detected. The Grubbs’ test p-values for this thesis are shown below in Table 7.4.254 In Appendix K, are shown estimates of the mean and standard deviation (sigma), in Fig. K.3 and Fig. K.4, which will be examined together with estimates designed to be resistant to outliers, shown below in Table 7.4.255

254 Grubbs’ test is not suitable for samples of six or less observations. 255 StatPoint Inc. (2006).

307

Table 7.4 Outliers diagnostics Hypothesis One Hypothesis Two Quarterly Monthly Weekly Daily Quarterly Monthly

Trimmed mean‡ -0.00061 2.12384 -0.82472 -0.19440 0.58428 -0.27537

Winsorised mean 0.00037 1.83076 -1.10106 -0.14799 -0.00327 -0.24357

Winsorised sigma 0.03453 25.8678 17.5436 6.53157 14.2055 3.34857

Lower limit† -0.01024 -8.84696 -4.43186 -0.69657 -4.3751 -1.6258

Upper limit† 0.01100 12.5085 2.22973 0.40057 4.36854 1.13865

Grubbs’ test 3.04919 2.92824 3.03493 4.43421 2.96044 2.40341

0.08944 0.06196 0.32402 0.00553 0.12587 0.43466 P-Values in italic. ‡ Trimming: 15.0%. † 95.0% confidence intervals for the mean (windsorised).

Using equation (19) as an example of how to read Table 7.4, together with other statistics available in Appendix K, Panel A in Fig. K.3, provides the values of the sample mean and sigma for residuals: 0.00000462 and 24.53, respectively. As shown in the table above, the corresponding Winsorised estimates, in which 15.0% of the largest and smallest data values are replaced by values from the interior of the sample, are 1.83076 and 25.8678, respectively.256 The impact of the Winsorised estimates on the confidence interval for the mean, -8.84 and 12.50, is therefore not significant. This finding extends to the remaining models.

Table K.1 and Table K.2 also exhibit other studentised values for residuals. Studentised values without deletion measure how many standard deviations each value is from the sample mean of the residuals. A graphical observation is made possible in Fig. K.3 and Fig. K.4. Continuing with equation (19) to illustrate the examination procedure, the most extreme residual is that in row 4, which is 2.92824 standard deviations from the mean (vid. Table K.1). Since the

256 StatPoint Inc. (2006).

308

P-value for Grubbs’ test is greater or equal to 0.05, that value is not a significant outlier at the 5% significance level, assuming that all the other values follow a normal distribution (vid. Table 7.4).

Confirming early indications, apart for the model using daily data, Grubbs’ p- -values are greater or equal to 0.05 for all remaining models. This means that Grubbs’ test did not consider that significant outliers exist at the 5% significance level, assuming that all the other values follow a normal distribution. In Table K.1 and Table K.2, are displayed similar scores for unusual residuals after deleting each point one at a time when computing the sample statistics, and when the mean and standard deviation are based on the median absolute deviation (MAD).257 Values the modified MAD Z-scores greater than 3.5 in absolute value may well be outliers. Again, only the model using daily data exhibits significant outliers, with five modified scores greater than 3.5. As the plot in Fig. K.3, Panel C, also points to the existence of several outliers for the model using daily data, some of which exceeding the extreme 4 sigma cutting-point, the possible effects of such outliers are to be studied in the next section of this thesis. influential points: High influential points are data points whose inclusion or exclusion result in significant changes in the fitted model (Makridakis et al., 1998). Although not so mediatised as outliers, high influential points have an impact on the regression analysis that can prove enough to shift the regression direction.258 Due to this possible influence power on the regression result, it is therefore relevant to identify unusual data that might have skewed the regression results obtained in this thesis.

257 StatPoint Inc. (2006). Vide modified MAD Z-scores in Table K.1 and Table K.2. 258 The distinction between outlier and influential point is not clear all times, as sometimes influential points are indistinctly mentioned in literature as being outliers. Nevertheless, since there is literature which separates both concepts, this thesis also makes the due distinction between concepts and related tests, even if such differentiation can be considered arbitrary, at times, by some authors.

309

There are several measures for detecting influential data points, being Cook’s D and DFITS some of the most commonly used.259 These measures combine information on residuals and leverage. Although they scale differently, they do provide somewhat similar results.

Cook’s D test compares all regression coefficients simultaneously. The lowest value that Cook’s D can assume is zero. The higher Cook’s D value is, the more influential the observation is. Montgomery et al. (2001), suggest that an observation with a Cook’s D greater than 1 can be a potential high influence point. For large samples, values greater than 4/n, being n the total number of observations, can be considered as influential.

DFITS is a statistic which measures how much the estimated coefficients would change if each observation would be removed from the data set (StatPoint Inc., 2006). In terms of rule of thumb, an observation can be considered to be influential if DFITS values are greater than ±1, in case of small-medium datasets. Unlike Cook’s D, DFITS can be either positive or negative. Numbers close to zero represent points with none or small influence. As for Cook’s D, this general rule of thumb depends on the number of observations used. A broader conventional cut-off point for DFITS is 2*sqrt(k/n), being k the number of predictors (Belsley et al., 1980).

Other measures include leverage and the Mahalanobis distance, which are used together with DFITS. According to StatPoint Inc. (2006), leverage is a statistic which measures how influential each observation is in determining the coefficients of the estimated model. As a general rule of thumb, a point with a leverage value greater than (2k+2)/n should be examined. The Mahalanobis distance is related to leverage. It measures the distance of a data point from the centroid of a multivariate dataset which is n-1 times the leverage of that data point.

259 DFITS is an acronym for “Differences in the FITS of an examined model”. It is also referred to in some literature as DFFITS.

310

In Table K.3 and Table K.4 the diverse statistics allowing to measure and to identify possible influential points on regression coefficient estimates are shown. The tables list all observations which have leverage values greater than 3 times that of an average data point; and/or which have unusually large values of DFITS (StatPoint Inc., 2006).

For the model used to test hypothesis one and using monthly data, there are no data points with more than 3 times the average leverage, which is 0.257143. However, there are 6 data points with unusually large values of DFITS. Cook’s D values are smaller than 1, but greater than 0.111(1), which corresponds to the broader cut-off point of 4/n. For the model using weekly data, there is one data point with more than 5 times the average leverage, 0.0645161. There are also 9 data points with unusually large values of DFITS and Cook’s D. For the model using daily data, there are 49 data points with unusually large values of DFITS. Furthermore, there are 32 data points with more than 3 times the average leverage, 0.0153649, and 7 points with more than 5 times. The existence of such a large number of unusually outsized values requires some further examination to determine how much the model would change if they were not present. Finally, the model using quarterly data has 4 data points with unusually large values of DFITS and Cook’s D, one of which exceeding 3 times the average leverage.

Regarding the models used to test hypothesis two, there are no data points with more than 3 times the average leverage, although there are several data points with unusually large DFITS and Cook’s D values.

In resume, only a model includes outliers, but all datasets include influential points. Nevertheless, the model that raises more meaningful concerns is the one using daily data, as it not only includes several influential points, but also outliers as well. Since this thesis has adopted the view that data should be used as raw as possible, avoiding the elimination of unusual data points or complex data smoothing procedures, in face of the facts presented in this section, it

311 becomes necessary to examine whether the models would change if such outliers and influential points were not present. The influence of these data points will then be subject of examination in the next section, together with other sensitivity analyses of interest for this thesis.

7.7.4 Sensitivity analysis

The concept of sensitivity analysis has been evolving, but it can be said that it is mostly concerned with studying and measuring the effect of a given input in a given output of a mathematical model (Saltelli et al., 2004). It is often performed using regression techniques (Saltelli et al., 2004 :42), and its importance is demonstrated by its widespread use in the literature using mathematical modelling, including in M&A research (e.g. Louis, 2004; Weber, 2004).

The purpose of sensitivity analysis in this thesis is twofold. On the one hand, its rationale comes from the importance of performing tests to assess the importance of the event variables in different time and statistical significance frames. On the other hand, by the means of robustness checks, it aims to measure the possible influence of outliers and influential points on stepwise backward regression procedure, which may have resulted in skewed models and in a biased elimination of variables.

The sensitivity analyses are focused only on the model using daily data, because: i) the comprehensive set of diagnostic tests performed did not indicate any substantial possibility of misspecification for the remaining models; ii) apart from the model using daily data, there is a small number of outliers and a relatively low number of influential points, being this particularly true for the models using monthly data; and, iii) the daily data sample is the only one

312 allowing the construction of several subsamples sizeable enough to run regressions testing the existence of any affects in the very-short term.260 extreme observations: It has been mentioned earlier that the use of daily data results in additional outliers in the models, at least when comparing with models using larger aggregation periods, such as the case of models using weekly, monthly, and quarterly data. Obviously, the use of daily data increases the randomness and is more likely to generate extreme observations, therefore possibly reducing the quality of the statistics.

Despite all efforts made in order to fit the models as best as possible, it is not possible, however, to capture entirely the effects of abnormal situations. In some extreme conditions, they can even lead to a statistical validation failure.

Following the identification of abnormal observations, as shown previously in the previous section and in Appendix K, several outliers and influential points were removed from the data sets, and new regressions were run. An extract of one of the alternative regressions run is shown in Appendix L. Since the original regression has four residuals with sigma values higher than 4, they have been eliminated as shown in Fig. L.2 (the four observations removed are shown as crossed in the plot). This elimination proved not only effective in terms of cleaning data from outliers, as it also observed the generic regression model assumptions. Nonetheless, some high influential points remained, or emerged, in the model.261 As shown in Fig. L.1, despite the elimination of the 4 major outliers, the Box-and-Whisker plot continues to identify several extreme points which indicate the possible existence of outliers. However, the Box-and-Whisker

260 Although not shown in this thesis, several sensitivity analyses were performed for the remaining models, including influential points’ elimination, and data trimming. Such alternative analysis, which included different time frames of analysis for the 1994-2008’s sample, neither changed significantly any regressions results nor changed the event variables elimination status. 261 Some data points are simultaneously outliers and high influential points.

313 plot identified “mild” outliers only, reinforcing the unlikelihood of the existence of “extreme” outliers.

The removal of the most extreme observations was followed by further eliminations of influential points. The regression estimates and the tests result of such alternative regressions provided similar results, and did not change meaningfully the initial conclusions drawn from the original models. non-trading days and reduced trading days: In order to assess any possible improvement of the quality of the dataset used, a set of non-conforming situations, which were identified earlier in this thesis, were also subject to eliminations. Such eliminations will enable to simulate whether the event variables could be statistically significant under different datasets arrangements.

In the early stage of the models’ development, M&A deals announced during weekends were subject to elimination from data series used in the models using daily data, because weekends are non-trading days. This elimination was also convenient from a statistic point of view, because it allowed addressing the issue of autocorrelation in the residuals, which was detected when variables including weekend data were used in seminal regressions.

Other non-trading days occurring during weekdays, resulting from holidays and extraordinary events, were left in the models, being treated with dummy variables. A similar treatment was given to reduced trading days, like the case of extended weekends. However, in spite of these situations being treated by using dummy variables, it is not possible to assure that the correspondent effects were captured, and therefore several subsamples were prepared to test the possible impact caused by the presence of these observations in the regressions.

314

Table 7.5 Resume of samples for sensitivity analysis

Sample Description Number of deals Number of days

2000-2002 completed M&A deals 19,758 1,096 Less Weekends (481) (314) Non-trading days (ordinary) (207) (26) Non-trading days (9/11 WTC) (25) (4) Reduced trading days (63) (8) Holiday seasons (36) (812) (3) (355) Smallest sample for sensitivity analysis 18,946 741

Résumé of samples Sample before sensitivity analysis 19,277 782 Removed Non-trading days (ordinary) 19,070 756 Non-trading days (abnormal) 19,045 752 Reduced trading days 18,982 744 Holiday season effects 18,946 741

Source: SDC Platinum (Thomson Financial, 2006).

As shown in Table 7.5, four subsamples were arranged from the original sample of 19,722 M&A deals and 782 days. The smallest sample does not include any non-trading, abnormal, or reduced trading day, and it is composed by 18,946 deals and 741 days. The regression model estimations and tests results are shown in Table L.1. The adjusted R-squared value is 51.03%, which compares with the original model value of 57.08%. Nonetheless, unlike the original model, all autocorrelations and partial autocorrelations values are within the 95% probability limits.262

262 Vid. Table L.2 and Table L.3, and also Fig. L.3 for a visual examination.

315

Table 7.6 Regressions’ sensitivity to abnormal and non-trading day’s removals Eliminations Original Non-trad. Non-trad. Reduced Holiday. Sample Ord. days 9/11 trading seasons Number of days 782 756 752 744 741

Adjusted R2 57.08 53.60 52.73 51.37 51.03

Durbin-Watson D 2.006 2.052 2.049 2.031 2.035 P-value 0.463 0.237 0.249 0.333 0.315

No. of variables entered 15 14 14 14 13

No. of variables selected 11 10 10 10 9

The results of the stepwise regressions with backward selection employing the subsamples exhibited in Table 7.5 are shown in the table above. The elimination of sets of observations from the original sample resulted in a decrease in the number of dummy variables possible to be utilised: following the elimination of ordinary non-trading days it was not possible to use anymore the variable Hol; and following the remaining eliminations it was no longer possible to use the HS_Ext variable. As the number of observations used was decreasing, the adjusted R-squared and the number of variables selected by the backward elimination procedure were also being reduced. However, the reduction in adjusted R-squared was not significant, and the variables left in the model by the backward elimination remained basically identical for all subsets tested. Although statistics for the regressions run are not shown in detail, one can ensure that the basic regression assumptions were observed.

The results obtained proved that the dummy variables that have been constructed to capture seasonality and abnormal events worked well, as they seem to have reasonably captured the pattern of M&A activity. In fact, not only the use of dummy variables increased the predictive power of the models, as its use did not result in biased regression results, since the regression model assumptions were observed. This finding reinforces the validity of the event

316 variables elimination, corroborating the non-rejection of the research hypotheses in the null form. alternative event windows: The use of two samples covering different sets of data, from 1994 to 2008, and from 2000 to 2002, was justified earlier in this thesis. Regarding the examination of possible short term impacts, the definition of a period of study of three years was discretionary, even if supported by reasoning, as mentioned before in this thesis. Therefore, it can be recognised that the selection of this event window of [-18; +18] months may be regarded as somewhat arbitrary.263 Moreover, there is always the possibility that the selection of other periods of study could result in different findings. Therefore, using the original sample for the model using daily data, other event windows were also controlled, covering shorter periods, namely: [-12; +12] months, [-6; +6] months, [-3; +3] months, [-1; +1] months, [-2; +2] weeks, and [-1; +1] weeks.

Table 7.7 Sensitivity analysis using alternative event windows [-12; +12] [-6; +6] [-3; +3] [-1; +1] [-2; +2] [-1; +1] Event windows months months months months weeks weeks Number of days 520 261 130 43 20 10

Adjusted R2 53.94 51.57 66.12 54.39 69.06 96.14

Durbin-Watson D 1.970 2.031 2.008 1.929 1.999 2.788 P-value 0.366 0.601 0.519 0.419 0.265 0.359

No. of variables entered 15 15 14 13 13 8

No. of variables selected 11 8 7 4 76

Although not shown in Table 7.7, the regression results for the alternative regressions covering shorter event windows, continued to exclude the

263 This is why a larger period of 15 years was also examined, as discussed before.

317 intervention variables. The adjusted r-squared values continue to be significant, particularly in very short-term event windows.

The systematic elimination of intervention variables in different testing contexts may be justified by the fact that two of the most abnormal days are not included in the samples for the model using daily data. As examined in chapter 4, the Saturday before the effectiveness date of the new M&A accounting standards clearly recorded an abnormal level of M&A activity, which would only be surpassed in dimension by the immediately following Sunday, the first day of the new accounting rules. Since weekends were eliminated from datasets, it would be therefore relevant to check whether the Event variable could be selected by the regression models, in case the 1 July 2001 weekend would be considered. Indeed, one could expect that, at least in short-term event windows, the inclusion of the event weekend could have a significant predictive power.

Similarly to the event windows shown in Table 7.7, other datasets were prepared including weekends, assuming market variables constant values during weekends. The regression results for the shortest event window examined, [-6; +6] days, included for the first time the Event variable, with a P-value of 0.0441.264 However, in larger periods, the intervention variables continued to be eliminated. Therefore, it is possible to conclude that the impact of the new business combinations accounting rules on the M&A activity was weakly perceived only at the micro-term level, at best. alternative significance levels: Every result shown before was produced using the backward regression with variables elimination at the 0.05 level. In this final sensitivity analysis stage, it is verified whether any event variable could be considered using different thresholds, namely at a tighter 99%, and at a stretched 90% confidence level.

264 Similarly to the methodology used by Branch et al. (2001), which added the M&A activity occurred during the weekends to the immediately following trading day, alternative regressions were run. Despite the adjusted datasets included 95 deals in the first post-event day, 2 July 2001, the event variables continued to be eliminated in very short-term regressions.

318

When the confidence level was raised to 99%, every event variable was eliminated, regardless the context of sensitivity analysis examined in this thesis. For the 90% confidence level, i.e., when the level of confidence defined for a variable to enter in the final model was reduced to 90%, it was registered a natural increase in the number of variables selected. Nevertheless, the event variables continued to be excluded, except in a few rare occasions.

The finding that even at a relaxed significance level the intervention variables continued to be eliminated in most situations, reinforces the previous finding that the accounting changes did not produce a significant impact on the M&A activity.

7.7.5 Forecasting model validation

In tandem with the diagnosis performed in the multivariate analysis section, it is also relevant to assess the appropriateness of the regression models, by means of measuring its forecasting accuracy. If a regression model is adequately fitted, and complies with the assumptions of regression with time series, but has a low predictive power, it will not be, therefore, of great usefulness. This is the main reason why it is pertinent to validate the models from a forecasting point of view. In the present research, this validation is to be done exclusively in the model using daily data, since this model is the one which uses the largest number of observations. Furthermore, this is the only model that uses a dataset including outliers, and also significant number of influential points, deserving therefore more attention, as pinpointed before.

In forecasting, accuracy often means goodness-of-fit, and it refers to which extent the regression model can reproduce the data already known (Makridakis et al., 1998). Being the forecast error the difference between an observation for time period t and the forecast for the same period, there are a number of standard statistical measures available for its analysis. It is the case of measures

319 such as the mean error (ME), the mean absolute error, the mean squared error (MSE), the mean percentage error (MPE), and the mean absolute percentage error (MAPE).

The Theil’s U inequality index is a statistical measure largely used for assessing the forecasting quality. It measures the degree to which one time series differs from another. In the present case, it compares the forecasts of the model using daily data with a naïve forecast, more precisely a random walk model. The U statistic ranges can be interpreted in a simple way. Whenever U equals one it means the forecasting power for the model is as good as for the naïve model. For values greater than one, it suggests that the fitted model forecasts are useless, since the use of a simple naïve model would produce a better forecast. Finally, if the U value is smaller than one, it means the model produces a relatively good forecast, as its predictions are better when compared to the ones that would be obtained using a naïve procedure.265

With the purpose of validating the model using daily data, the accuracy of an out-of-sample forecast has been measured. Following the procedure as described in Makridakis et al. (1998: 46), the data was divided into an “initialization” set and into a “holdout” set, corresponding to the last fifty observations. The parameters were estimated from the initialisation set, while forecasts were made for the holdout set, or “testing” set. As the holdout set does not belong to the fitted model, the forecasts produced are genuine, since they do not use any of the real observations. In order to produce forecasts, fifty regressions were run, using t-1 to t-50 observations, and a holdout set of fifty t+1 predicted values, one for each regression, was assembled. Finally, the predicted values were compared to the real observations to have the “real” forecast errors. Every accuracy measure was computed for the errors in the holdout set only, being exhibited below in Table 7.8.

265 The smaller the U value, the better the forecasting technique in relation to the naïve method (Makridakis et al., 1998: 50). For more information about Theil’s U statistic see e.g. Makridakis et al. (1998), or Theil (1966) himself.

320

Table 7.8 Out-of-sample accuracy measurement Model ME MAE MSE MPE MAPE

Daily data -1.48576 5.433921 54.11527 -17.84% 33.31%

Naïve 0.38 6.74 65.66 -28.88% 61.10%

Theil's U 0.593949 Measures computed using a holdout set of 50 observations.

From the comparison of the model developed to test hypothesis one with the random walk method, one can conclude that the model using daily data presents better statistical measures and a much better predictive power. In fact, although the naïve model exhibits a mean error closer to zero, all the remaining standard statistical measures of forecasting errors, MAE, MSE, MPE, and MAPE, present lower values for the model of equation (21). Since these measures are the most relevant ones, one can firmly argue that the model using daily data has a much lower overall forecast error than the naïve model.

This finding is corroborated by the Theil’s U statistic value, 0.59, which is much lower than one, consequently meaning that the forecasting power of the model using daily data is clearly superior to the one of the naïve model. In resume, from the analysis of the forecast errors, together with the Theil’s U inequality index, it can be concluded that the model using daily data has a reasonably good predictive power.

7.8 Conclusions

Following the development of the research hypotheses in chapter 4, and the presentation of the sample datasets in chapter 6, the main research testing was conducted in this chapter, by means of stepwise backward regression with time series. The results obtained were subject to in-depth robustness checks and statistic validations in order to minimise the possibility of comprising significant bias. Overall, the tests did not reveal problems regarding serial correlation and

321 heteroscedasticty. Even if some departures from normality were identified, as shown by statistics such as skewness, they did not violate the overall regression analysis assumptions. Such departures from normality could have been corrected through the elimination of outliers and influential points. However, since the inclusion of outliers in the models did not violate the regression assumptions and did not produce any material changes in the regression estimations, it was decided to keep the original data unchanged, relying in dummy variables to capture as much as possible any abnormal effects.

The main finding of this chapter is that the intervention variables were not selected by any model, to the detriment of other economic-financial and time variables, suggesting that the changes in business combinations accounting did not produce a significant impact on M&A activity, both in the medium and long terms. Therefore, the research hypotheses could not be rejected in the null form. The evidence obtained from the diverse sensitivity analyses conducted corroborated the non-rejection of the hypotheses, also suggesting the inexistence of impacts in very-short and in the short run.

The only significant evidence contrary to the main finding was found in sensitivity analysis, where Event variable was selected for one model, but only in a particular situation and using a very-short event window of [-6; +6] days. The discussion started earlier in this thesis about collateral effects resulting from the M&A accounting changes follows in the next chapter.

322

Chapter 8 Discussion and Interpretation

8.1 Introduction

This thesis relates to different strands of research from accounting, finance, and economics. More precisely, this research examines several issues in financial reporting, namely the ones concerning accounting regulation and accounting choice, extant with lobbying, as discussed earlier. Of particular interest is the issue of accounting choice, which has been tested in order to examine whether the constraint placed in business combinations accounting produced significant economic consequences, namely on M&A activity.

In the context of accounting choice in business combinations, two main issues were examined: i) pooling of interests versus purchase method, which has been a perennial source of controversy until recently, as evidenced by the sizeable amount of related literature produced in the past decades; and, ii) impairment testing versus amortisation of purchased goodwill and other intangible assets, which became a focus of discussion in business combinations after pooling of interests option was discontinued by the FASB, a move that would also be followed soon afterwards by the IASB.

The main issues of accounting choice in business combinations accounting were therefore examined in this thesis, although not necessarily fully tested, as discussed in chapter 5. The interest and usefulness of further testing is to be

323 discussed in this chapter, in connection with some final remarks to be made in the concluding chapter.

Furthermore, this thesis also deals with one of the major issues in finance research, which is the absence of a comprehensive theory explaining the occurrence of M&A waves. Although not particularly concerned with M&A waves itself, this research is related to the pattern of the M&A activity, and therefore to diverse strands of literature in M&A, like research on M&A completion and cancelations, also being relevant the contributes from behavioural finance. Concurrently, this thesis is also related to economics, as diverse modelling developments on M&A’s activity and waves come from this area of research. Additionally, this thesis uses macro and microeconomic factors, together with other exogenous and endogenous factors, in order to model the M&A activity according to the purposes drawn in the research hypotheses chapter.

The development of specific models to test the hypotheses designed for this thesis has been undertaken in the previous chapter, which has also examined the results obtained. As for any testing, there is always room for some further adjustments and improvements. They are therefore to be examined in this chapter, being the limitations discussed in chapter 9.

Finally, the globalisation of markets and financial reporting requires any comprehensive research in accounting, finance, or economics, to include also a global angle. Accordingly, this chapter is particularly focused on relating the research findings to different international scenarios, with particular emphasis to Europe, and to other countries using or considering adopting IASB’s standards.

324

8.2 Accounting regulation and accounting choice: an international perspective for business combinations

As discussed earlier, the reactions to FASB’s proposals on new business combinations accounting suggested that its adoption could constrain M&A announcements and affect the completion of M&A deals. The opposition to the first exposure draft can be regarded as normal, as FASB simply wanted to eliminate the accounting choice in business combinations, by discontinuing pooling of interests. Moreover, the FASB proposed reducing the maximum period amortisation of goodwill and other intangible assets from 40 to 20 years. This early proposal, that would most certainly affect companies using or intending to qualify for pooling of interests, was particularly penalising for companies engaged in M&A deals involving substantial values of goodwill and other acquired intangible values. It could be even regarded as a double penalisation. On the one hand, with the pooling option elimination, every company involved in a M&A deal with an acquisition price superior to the total value of assets and liabilities measured at fair value, would now be necessarily forced to recognise goodwill. On the other, as shown in chapter 2, since the majority of companies amortising purchased goodwill were doing so in around 40 years, a substantial number of companies would now have to record higher amortisation charges, consequently affecting negatively their reported earnings.

These earlier proposals were regarded as a menace for companies that were using pooling to avoid recognising goodwill whenever accounting for M&A deals. Opposition from relevant sectors of activity was therefore inevitable, and could ultimately jeopardise the proposed changes in business combinations accounting. However, besides backing its proposals arguing from a technical standpoint - supported by feedback obtained from invitations to comment and public hearings - the FASB had another powerful argument on its side: the need for an international framework for accounting extant with globalisation of business and markets. It is important to recall that the FASB compromised and worked together with other standard-setting bodies worldwide in order to achieve international convergence on the methods of accounting for business

325 combinations (vid. e.g. Paragraph B9, SFAS 142, Financial Accounting Standards Board, 2001e; G4+1 (Organization) & International Accounting Standards Committee, 1998). In fact, it is important to recall that the works on a harmonised accounting for business combinations were made under the auspices of G4+1 Organization, having the IASB as a special observer.

Unsurprisingly, as a way to legitimise and reinforce its standard-settings options, the FASB openly claimed the need for international convergence in business combinations accounting. This was made clear in official statements, such as in SFAS 142 (Paragraph B49, Financial Accounting Standards Board, 2001e):

“The Board observed that Opinion 17 required intangible assets to be amortized over their expected useful lives; however, amortization periods were limited to 40 years. The Board noted that standards elsewhere that address intangible assets are generally similar. However, in some cases, the maximum amortization period is less than 40 years, with 20 years frequently being the presumed or absolute maximum.”.

The FASB inclusively did refer to accounting standards from other standard- -setting boards, such as the UK ASB, or the international IASC, in order to illustrate and to support its basis for conclusions on M&A GAAP’s changes (SFAS 142, Paragraph B49, Financial Accounting Standards Board, 2001e):

“The Board noted that both FRS 10 and IAS 38 have presumptive maximums of 20 years. However, FRS 10 permits some intangible assets not to be amortized at all, provided that (a) the durability of the asset can be demonstrated and justifies an amortization period longer than 20 years and (b) the asset is capable of continued measurement so that annual impairment reviews can be conducted. IAS 38 requires all intangible assets to be amortized but does not specify a maximum amortization period.”.

As discussed earlier in this thesis, following a period of intense lobbying, including not only corporate lobbying but also congressional pressure, the FASB

326 would eventually drop the proposal of amortisation of purchased goodwill and other intangible assets.266 However, FASB’s position on pooling of interests method elimination would not change. As pooling method permission was discontinued in the USA, the path for an international convergence in business combinations accounting was finally set.

Table 8.1 Accounting for business combinations worldwide in the 1990’s Pooling method if Purchase method if uniting of interests uniting of interests Australia Not permitted Required Brazil Not permitted Required Canada Required Not permitted France Permitted Preferred Germany Permitted Permitted Italy Permitted Permitted Japan Not permitted Required Netherlands Permitted Permitted Portugal Permitted Permitted Spain Permitted Permitted Sweden Required Not permitted Switzerland Permitted Permitted United Kingdom Required Not permitted United States Required Not permitted Source: adapted from Radebaugh & Gray (1997: 254).

As shown in Table 8.1, the options concerning the pooling of interests utilisation were very much heterogeneous at worldwide level by the mid-1990’s. Most countries allowed the use of both pooling and purchase methods in case the M&A deal could be qualified for a uniting of interests, in spite of pooling of interests being considered as the most appropriate method for a uniting of ownership interests of two or more companies, by means of exchange of equity

266 See e.g. Beresford (2001) and Zeff (2002).

327 securities. However, in countries such as Australia, Brazil, and Japan, it was not allowed to account for uniting of interests using pooling of interests, as it was mandatory to use purchase method instead. Conversely, in Canada, Sweden, the UK, and in the USA, pooling of interests was mandatory for uniting of interests.267 Finally, in France the purchase method was preferred for uniting of interests, but the use of pooling of interests was also permitted.

The need for convergence in business combinations accounting was becoming increasingly obvious. In a time distinguished by increasingly unrestrained globalisation, financial reporting has to provide information for international users (e.g. Ball, 2006). For globalised users, such financial information has to be internationally intelligible and preferably harmonised, in order to be effectively

267 While pooling of interests method was required for uniting of interests in the USA, and in some other countries, purchase method was required for all remaining business combinations. Therefore one could argue that accounting choice in business combinations did not exist in these countries. Indeed, AICPA’s APB stated in APB Opinion 16 that (Paragraph 8, American Institute of Certified Public Accountants; Accounting Principles Board, 1970b):

“The Board concludes that the purchase method and the pooling of interests method are both acceptable in accounting for business combinations, although not as alternatives in accounting for the same business combination. A business combination which meets specified conditions requires accounting by the pooling of interests method. A new basis of accounting is not permitted for a combination that meets the specified conditions, and the assets and liabilities of the combining companies are combined at their recorded amounts. All other business combinations should be accounted for as an acquisition of one or more companies by a corporation.”.

However, as discussed in earlier chapters, companies managed to qualify M&A deals to uniting of interests, distorting APB Opinion 16 intentions (see e.g. Aboody et al., 2000; Ayers et al., 2002; Linsmeier et al., 1998; Lys & Vincent, 1995; Robinson & Shane, 1990; Walter, 1999; Weber, 2004). In fact, often pooling and purchase methods were effectively “alternatives in accounting for the same business combination”, and therefore accounting choice was no less than a common reality in M&A accounting. So common that could be inclusively purchased, as illustrated by the acquisition of NCR by AT&T, were it is suggested by Lys & Vincent (1995) that AT&T incurred in at least $50 million of additional costs to qualify the deal as a uniting of interests (vid. chapter 2).

328 useful. Additionally, transnational companies were also interested in reducing the costs of preparing financial information. Therefore, international accounting convergence was critical in order to reduce the need for reporting differently according to different domestic GAAP sets, an undesirable situation often resulting in the need for numerous GAAP reconciliations by transnational companies.

One example, which became a “classic” case study in international accounting convergence was Daimler-Benz. Due to strategic, but also financial reasons, Daimler-Benz became the first German company to be listed in the NYSE in 1993. However, this dual listing would also result in a dazzling dual financial reporting. In the first year reporting according to the US GAAP, by means of reconciliation, Daimler-Benz revealed a translated net loss of 1,839 million Deutsch Marks (DM), while concurrently it reported a profit of 615 millions DM under German GAAP (L. H. Radebaugh et al., 1995). This situation would continue to be repeated, as Daimler-Benz profits under German GAAP were systematically reduced following reconciliation with the US GAAP (Roberts et al., 1998). However, in 1995, when Daimler-Benz reported a net loss for investors in Frankfurt a curious reversal would occur, as barely the same figure would be reported for investors in New York - but a net profit instead. The differences of reconciliations from German and the UK GAAP to the USA GAAP were large in some occasions, including steep variations from year to year, puzzling accounting professionals and users of financial statements (Alexander & Nobes, 2001). In 1995, Daimler-Benz ceased the reconciliation of financial statements as it ultimately decided to adopt the USA GAAP (Christopher Nobes & Parker, 2002). As situations like the one of Daimler were multiplying among multinationals, the overall reliability and usefulness of financial reporting was being increasingly jeopardised, and companies were therefore urged to enhance international accounting convergence.268

The use of international financial reporting standards also carries some financial benefits (Ball, 2006). Indeed, while increasing the harmonisation of the

268 For overall international convergence of accounting practices see e.g. Tarca (2004).

329 interpretation of financial reporting worldwide, the analysis costs for international users are also reduced. Similarly, on the preparers’ side, it also reduces the costs for multinational companies, as they no longer need to prepare different sets of financial statements under different accounting frameworks. It also reduces the information risks for investors (Ball, 2006).

Literature has increasingly been examining the adoption of international financial reporting standards. For example, Barth et al. (2008), while examining companies using IASB’s accounting standards in 21 countries, found less evidence of earnings management, more timely losses recognition, and more value relevance, than in sampled companies applying the non-USA domestic GAAP, in between the pre- and post-adoption periods.

Furthermore, in many cases the adoption of the - in general - less conservative IASB’s GAAP leads to an increase in firms’ value. The extended use of fair value by IASB is an example (e.g. Ball, 2006). Indeed Karamanou & Nishiotis (2009), using a sample of international firms, found the existence of abnormal returns at the announcement of voluntary adoption of international accounting standards, therefore suggesting that companies can increase their market values through their choice of accounting standards.

Nevertheless, there are not only advantages and positive factors. Indeed, international harmonisation of accounting standards has both considerable pros and cons. For example, Ball (2006) refers pertinent cons, such as the fact that IASB pushed fair value accounting “too far”. The use of fair value carries some issues, such as an emphasis on “relevance” to the detriment of “reliability”, or the downplaying of the indirect “stewardship” role of accounting (Ball, 2006). Other cons include the fact that politics and markets continue to remain primarily local; or even concerns with the long-term undermining of IASB by politics and bureaucracy (Ball, 2006). Indeed, IASB’s standards implementation raises concerns related to its uniformity when facing incentives of financial reporting preparers and enforcers, such as auditors, courts, regulators, politicians, etc., which remain primarily local (Ball, 2006). Similarly, being

330 aware of the lobbying and political pressures surrounding IASB (vid. e.g. Chand & Cummings, 2008; Zeff, 2002), it seems reasonable to consider the risk of its increasingly transformation in a representative, polarised, and bureaucratic body (Ball, 2006).

Finally, as discussed in this thesis, the accounting choice in business combinations can lead to considerable changes in financial reporting, exacerbating potential reporting disadvantages for international companies that do not have alternative M&A accounting methods available. Taking again the Daimler-Benz example, this firm was originally listed in Frankfurt, but in order to prepare its expansion in the USA it became listed in New York as well. Indeed, it would later integrate the North American Chrysler. Following the acquisition of Chrysler Corp., in 1998, a new company called DaimlerChrysler was formed, by means of a formal merger, which managed to qualify for pooling of interests method (Coleman, 1998). This was the case of a deal where Daimler-Benz was de jure merging, but de facto acquiring Chrysler, reshaping also the transaction in order to qualify contra natura for pooling.

Following G4+1 proposals on business combinations, and subsequently with the adoption of the new accounting standards in the USA, it was urged for IASC/IASB and other accounting standard-setters worldwide to move alongside with the due convergence. It was also a critical time for Europe, particularly for the EU Member Countries, as the possible adoption of IASC/IASB’s standards was being discussed.

The number of different accounting practices at national level is considerable in Europe, and the accounting for business combinations provides a good example of a profusion of different methods that may be possibly used by companies in different European countries, with implications on the comparability of financial statements (Fédération des Experts Comptables Européens (FEE), 2002; Ribeiro & Crowther, 2008a).

331

Table 8.2 Accounting for M&A in Europe and in the USA: 1999-2000 Pooling Goodwill accounting treatment Conditionally Write Off to Amortisation Period Limits Possible Equity Allowed Rebuttable† Absolute† Austria No Yes - None Belgium Yes - 5 None Czech Republic No Yes - 5 Denmark Yes Yes 5 None Finland Yes - 5 20 France Yes Yes (20) None (40) Germany Yes Yes 20 None Hungary No - 5 15 Ireland Yes - 20 None Italy Yes Yes 5 None (20) Luxembourg Yes Yes - None Netherlands Yes Yes 5 None Norway Yes - 5 (20) None Poland Yes - 5 20 Portugal Yes - 5 20 Romania Yes - - 20 Slovenia Yes - - 5 Spain Yes - - 20 Sweden Yes - 5 20 Switzerland Yes Yes 20 None United Kingdom Yes - 20 None United States Yes - - 40 Source: adapted from Fédération des Experts Comptables Européens (2002: 21, 27). † Number of years. (In brackets: there are no formal limits, but tacitly amortisation should not exceed).

By the time the USA was discussing the changes in business combinations, the plethora of different accounting methods and practices for business

332 combinations in Europe was noteworthy. From the twenty-one European countries studied by the FEE in 1999-2000, only three did not allow the use of pooling of interests in any circumstance, so it can be argued that there was some degree of convergence regarding the allowance of alternative business combinations methods. However, in terms of purchased goodwill accounting, there were significant differences among countries. Despite amortisation being the preferred treatment, eight countries allowed write off to equity. Additionally, fifteen countries had refutable amortisation period limits, and ten countries had no absolute limits for goodwill amortisation. Furthermore, the periods of amortisation for the countries with absolute limits varied substantially.

As mentioned before in this chapter, and as discussed earlier in this thesis, the IASB followed closely the FASB’s business combinations developments. Later, both boards worked together in this subject matter, in the scope of a convergence project. Accordingly, the IASB would publish IFRS 3 afterwards, in 2004, adopting also the principle that the purchase method should be applied to all business combinations. The IASB also shared FASB’s views on purchased goodwill, arguing that it should not be amortised, and determining that any recognised goodwill should be carried at cost less any accumulated impairment losses, as required by IAS 36 (International Accounting Standards Board, 2004).

In 2005, IASB’s standards would be enforced by the EU. If the accounting for business combinations in Europe was already a Tower of Babel, another layer of diversity would be then added following IAS/IFRS adoption by the EU countries (Fédération des Experts Comptables Européens (FEE), 2002; Ribeiro & Crowther, 2008a).269 The adoption of international accounting standards was a relevant step towards accounting convergence in Europe, but it has also resulted in a dual accounting, with its inherent inconveniences.

Indeed, the adoption of IASB’s standards resulted in a dual accounting system, in which any listed company is compelled to prepare consolidated accounts

269 The remainder of this section is based in Ribeiro & Crowther (2008c, 2008a).

333 according to IASB’s standards (European Commission, 2002), while the remaining firms are allowed to keep using national sets of accounting standards, which are based on the aged fourth and seventh EU directives.

More precisely, within this dual accounting system, since 2005 that listed companies ruled by the law of a Member State are required to prepare their consolidated accounts according to IASB’s standards (European Commission, 2002). The matrix of possibilities has increased. Listed companies with consolidated accounts can only use purchase method, but pooling of interests continues to be allowed for the remaining companies, except in a few countries. Similarly, for companies using domestic GAAP, purchased goodwill is to be amortised, using arbitrary ceilings; or is to be written off; while for companies adopting IASB GAAP, goodwill is to be subject to impairment tests. As shown before, the discrepancy among the different national sets of GAAP in Europe is significant, and, paradoxically, the asymmetries within the European zone have been reinforced with the introduction of IASB’s GAAP.

The UK is a good example of these dual accounting discrepancies. Currently, business combinations in the UK are accounted according to Financial Reporting Standard (FRS) No. 6, Acquisitions and Mergers (Accounting Standards Board, 1994a), FRS 7, Fair values in Acquisition Accounting (Accounting Standards Board, 1994b), FRS 10, Goodwill and Intangible Assets (Accounting Standards Board, 1997), and FRS 11, Impairment of Fixed Assets and Goodwill (Accounting Standards Board, 1998), which are not in conformity with IFRS 3, as pooling is allowed, and goodwill is to be amortised. The ASB keeps pursuing convergence with IASB’s GAAP, and has published several financial reporting exposure drafts (FRED) in order to achieve this purpose.270

270 In order to achieve a broad convergence with IFRS 3, and related IASB’s standards on business combinations and goodwill accounting, like IAS 36, and IAS 38, the ASB issued several exposure drafts in 2005, FREDs 36, 37, and 38 (Accounting Standards Board. Financial Reporting Council, 2005a, 2005b, 2005c), which also include proposed amendments to FRS 2, Accounting for Subsidiary Undertakings (parts of IAS 27, Consolidated and Separate Financial Statements).

334

This means that business combinations accounting should be likely to change in the short term at the domestic level in the UK.

Nevertheless, as time passes uncertainty is rising, being the concretisation of the proposed changes for business combinations unforeseeable. Furthermore, this uncertainty covers the entire future reporting requirements for the UK and Irish entities. Indeed, following the issuance of a consultation paper in August 2009, the ASB announced to be discussing a possible triple accounting framework, (vid. http://www.frc.org.uk; Accounting Standards Board. Financial Reporting Council, 2009):

“The Board is proposing a three-tier approach to developing UK GAAP converged with IFRS as follows:

Tier 1 – publicly accountable entities would apply IFRS as adopted by the EU (‘EU-adopted’ IFRS).

Tier 2 – all other UK entities other than those who can apply the Financial Reporting Standard for Smaller Entities (FRSSE) could apply the IFRS for SMEs.

Tier 3 – small entities could choose to continue to apply the FRSSE.

Entities within Tier 2 and Tier 3 would have the option of using EU-adopted IFRS if they wished, and those in Tier 3 would have the option of using the IFRS for SMEs.”.

Is it the case that companies in the UK may benefit from a triple accounting system? This may not be the case for business combinations accounting, as smaller companies should not need to have an adjusted accounting for M&A deals. But as convergence with IASB is being put on hold, the dual accounting in business combinations continues to be a fact in the UK.

335

Since only a small number of companies are listed and have to prepare consolidated accounts, and as several national sets of accounting standards continue to allow the use of pooling, this means that a very considerable number of companies are still able to utilise pooling of interests in Europe, in contrast to a few large companies that have necessarily to use purchase method and to recognise purchased goodwill. This is a relevant fact, as it materialises one of the main contributions of this thesis, as to be discussed in the final chapter.

8.3 Accounting choice in business combinations accounting and M&A activity

As discussed earlier, before FASB’s changes in 2001, the managerial accounting preference in business combinations was evident: pooling of interests method was more wanted, regardless its use being conditional to its qualification as a uniting of interests (see e.g. Aboody et al., 2000; Anderson & Louderback III, 1975; Ayers et al., 2000, 2002; R. M. Copeland & Wojdak, 1969; Gagnon, 1967; Lys & Vincent, 1995; Nathan, 1988). Management seeks to maximise results, and purchase was not the best method for such aim, as it required goodwill recognition and its amortisation, with negative consequences on earnings. Indeed, for a long time that literature found that managerial discretion was used in business combinations accounting in order meet financial reporting objectives, namely to maximise reported earnings (R. M. Copeland & Wojdak, 1969).

However, this managerial appetence for pooling of interests was excessive. In fact, diverse anecdotal and empirical evidence suggested that companies could reshape M&A deals, incurring in extraordinary expenses, and even paying higher acquisition premiums, simply to meet APB Opinion 16 pooling of interests criteria (see e.g. Aboody et al., 2000; Ayers et al., 2002; M. L. Davis,

336

1990; Hopkins et al., 2000; Linsmeier et al., 1998; Lys & Vincent, 1995; Robinson & Shane, 1990; Walter, 1999; Weber, 2004).271

Despite the apparent advantages of pooling over purchase method, several studies detected that pooling method resulted in artificial arrangements on companies’ financial statements, distorting the corresponding financial analysis (Jennings et al., 1996; Vincent, 1997), while others documented that the announcement returns were lower for pooling companies when compared to companies using purchase method (M. L. Davis, 1990; Hong et al., 1978; Martinez-Jerez, 2001).

According to Fields et al. (2001), whether shareholders benefit from managerial discretion and whether the benefits outweighed the costs is not clear. However, according to Louis (2004), literature suggested that pooling deals are “bad investment decisions” because managers miss the focus on cash-flows, as they are more concerned with reporting increasing earnings, and also because they constrain the management’s ability to sell acquired assets after the acquisition (Lys & Vincent, 1995; Robinson & Shane, 1990). Unsurprisingly, Martinez-Jerez (2003) found that a stronger negative market reaction to pooling M&A deals is linked to acquirers that have poor corporate governance.

On the one hand, pooling seemed to underperform purchase method, but on the other hand, its defenders raised their arguments loud when the FASB proposed to eliminate pooling. Apparently, using the theoretical model of Watts & Zimmerman (1990, 1978) for the understanding of the developments on M&A accounting, among pooling supporters were certainly managers more concerned with their contracts, than with any economic consequences from the forthcoming changes in M&A accounting, as shown by the fierce lobbying undertaken (Zeff, 2002). Watts & Zimmerman (1978) suggested that accounting choice may affect shareholders’ wealth in case managers’ compensation contracts are coupled to financial reporting performance. Indeed, while

271 Conversely, Nathan (1988) did not find higher acquisition premiums for companies applying the pooling of interests method.

337 examining specific characteristics determining which business combinations accounting method is selected by the management, the literature found that the percentage of insiders’ ownership, and accounting-based compensation plans play a significant role (Dunne, 1990). In resume, it may be suggested that much of the lobbying surrounding the business combinations accounting changes may have been related to executive interests. Furthermore, such interests appear to have been safeguarded with the replacement of purchased goodwill amortisation by impairment tests. The financial industry constitutes a good example. On the one hand it developed a very active lobbying action on the FASB and on the Congress of the USA, but on the other it was one of the least impacted industries by the new accounting rules, with average values of purchased goodwill and other intangibles below the mean. Why to bother then, why to be more actively lobbying than other industries, such as consumer discretionary, or industrials, which have been much more impacted? Perhaps because besides the interest on the M&A fees that could be lost due to a reduction on the number of M&A deals, were also contractual relationships that were being jeopardised. research findings related with literature pre-changes in M&A accounting: The results obtained from testing the research hypotheses, as shown in chapter 7, corroborate the overall findings from literature on accounting choice in M&A pre-new business combinations accounting. As discussed, this strand of literature pointed out disadvantages in business combinations accounting choice, by means of examining the existence of pooling of interests as an alternative method. By suggesting that the changes in M&A accounting, namely the discontinuation of pooling of interests method, did not result in significant impacts on M&A activity, the findings of the present thesis also support the firm position of FASB in eliminating the accounting choice in business combinations accounting.

Furthermore, evidence collected from questionnaires and annual reports of S&P 500 companies also corroborates the findings of the literature existing before pooling of interests elimination. Some survey respondents expressed serious

338 concerns about the new standards. Additionally, evidence from S&P 500 companies’ annual reports made it possible to conclude that had SFAS 142 been made effective in 2000 and diluted EPS from S&P 500 companies would have increased 20.7% on average. Similarly, adoption of SFAS 142 in the fiscal year 2001 would produce an even more significant average increase of 29.6% in diluted EPS.

Some authors have anticipated or estimated similar impacts under different, but related circumstances. For example, Ayers et al. (2000), using a sample of pooling companies, estimated that EPS would have been considerably lower if purchase method had been used. By the time this study was made, purchased goodwill and other intangible assts were being amortised over a maximum period of 40 years.

The Ayers et al.’s paper is interesting because it offers a comparable measurement. In Ayers et al.’s study, assuming a 10-year amortisation period, the decrease in EPS would be from 8.3% in financial services, up to 42% in food, textile, and chemicals industries. Assuming a 40-year amortisation period, EPS figures would be reduced from 2.2% in financial services, up to 15.7%, in the hotel and other services industries.

As FASB dropped the initial proposal on replacing a 20-year amortisation period for impairment tests, it has reversed entirely the impact on earnings, at least in a scenario of absence of impairment losses. Comparing the initial FASB’s proposal with the final provisions of SFAS 142, this outcome has resulted in a reversal of possible losses of over 15% in some industries, to average increases in earnings of more than 20%.

The indications obtained in this thesis of a possible impact on M&A activity in the IT industry also corroborates literature and public concerns of the sector due to the proposed elimination, as discussed in earlier chapters (see e.g. King, 2000; King & Kelly, 2000; Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, 2000). These claims are

339 everything but overrated: the majority of respondents to the questionnaire from IT industry pinpointed the existence of effects from SFAS 141 and SFAS 142 adoption, and IT firms were also the most likely to be impacted by the new accounting rules among S&P 500 companies, as they were carrying the highest values of purchased goodwill in their balance sheets and could therefore be more impacted by the replacement of goodwill amortisation by impairment tests than any other sector of activity. research findings related with literature in finance and economics: The models developed in this thesis are related to diverse literature in finance and economics, namely to the one related to M&A activity and M&A waves. Nevertheless, it was not intended to draw conclusions related to any underlying theory concerned with explaining the occurrence of M&A in waves. However, it was needed to understand the pattern of M&A activity during the period in analysis in order to develop and adequately fit the models used to test the research hypotheses.

Despite carefully fitted, the models are not free from limitations. One of these limitations is the impossibility of extending the number of variables in the models, particularly in the cases whenever data is not produced for very-short periods, such as weekly or monthly. Another limitation is related with the absence of subsamples breaking down M&A data by industry. This limitation prevented performing cross-sectional analysis, which could refine the overall finding of absence of impacts from the new M&A accounting on the M&A activity. As examined before, there are indications that M&A activity in IT industry may have suffered from a certain degree of constraint. Consequently, it would be very interesting to have M&A subsamples by industry, in order proceed with hypotheses testing at industry-level. Despite these limitations - and others that are to be discussed in the concluding chapter -, and the obvious possibilities of further adjustments and improvements, there is confidence in the models and in the results obtained in this thesis.

340

Following the models testing, other findings include corroboration of diverse literature on the M&A pattern activity. Overall, the signs of the variables tested were consistent with most findings from literature, apart the case of interest rates - a divergence which has been justified in chapter 7. research findings related with literature post-changes in M&A accounting: Following the effectiveness of the new business combinations accounting rules, the debate shifted from pooling elimination to the accounting treatment of purchased goodwill and other intangible assets. Amortisation versus impairment testing became the main issue. As examined earlier, SFAS 142 replaced amortisation of purchased goodwill and other intangible assets with indefinite useful lives by impairment tests, keeping amortisation, under certain conditions, only for goodwill and intangible assets with finite useful lives. By doing so, FASB added volatility to the financial reporting. This fact was assumed in SFAS 142. When comparing the differences between SFAS 142 and APB Opinion 17, it is stated that (Financial Accounting Standards Board, 2001e: 5):

“Because goodwill and some intangible assets will no longer be amortized, the reported amounts of goodwill and intangible assets (as well as total assets) will not decrease at the same time and in the same manner as under previous standards. There may be more volatility in reported income than under previous standards because impairment losses are likely to occur irregularly and in varying amounts.”.

Certainly there will be more volatility, one could add. Regardless the merits of impairment treatment, which screens continuously for the value of purchased goodwill and other intangible assets, whether such volatility increases the quality of financial reporting, and the usefulness of its users, is not a clear matter yet. Perhaps the best judgement will come with time, but in the meantime some literature started to examine this issue, providing some early findings.

341

In terms of examples of literature about impairments under SFAS 142, Hayn & Hughes (2006) questioned the superiority of impairment tests over amortisations, while Massoud & Raiborn (2003) suggested that the managerial discretion in applying the goodwill impairment tests reduces the quality of reported earnings. Massoud & Raiborn (2003) argue that SFAS No. 142 creates opportunities for earnings management, particularly in weak economic periods, where companies can undertake a “big bath”, i.e., to recognise big impairment losses in a period when earnings are already negatively affected. According to the analysis made in this thesis of annual reports from S&P 500 companies, this was indeed the case in 2002, as a significant number of companies’ recognised outstanding values of impairment losses under SFAS 142, therefore supporting the “big bath” earnings theory.

8.4 New M&A accounting rules and M&A activity: business as usual?

The evidence shown in this thesis suggests that the new business combinations accounting rules may have had only a limited impact on M&A activity, at very best. Following the new FASB’s standards implementation, in terms of the number of companies accounting for business combinations, the figures remained stable, despite the pooling elimination and the changes in the purchase method. For example, according to the Accounting Trends & Techniques (American Institute of Certified Public Accountants, 1968-2003), in 2002, 314 of the 600 surveyed companies used the new purchase method to account for a business combination, versus 309 purchases and 16 pooling of interests in 2001.272

Additionally, there were not indications of any relevant economic effects produced by the new standards. While the neutrality of the new standards seems not to be an issue, not everything remained necessarily unchanged. Otherwise, FASB’s changes would not be more than sheer cosmetic and would

272 It is important to recall that, in 2001, pooling of interests was possible only until 30 June.

342 not produce any effects. Such outcome could be, perhaps, considered even to be abnormal. As Hendriksen & van Breda (1992: 242-243) note:

Therefore, the appropriateness of the changes in business combinations and goodwill accounting may be perhaps simply judged by the absence of significant effects. Conversely, one could argue that the accounting changes should be useful, resulting in positive effects for the users, particularly investors.

Added transparency brought by the new rules also seems an unquestionable fact, as financial reporting discrepancies between pooling versus non-pooling firms disappeared. This should have contributed to reduce the cost of financial reporting analysis, and may have also increased firms’ value. In the past, analysts placed lower valuations on companies that amortised goodwill using the purchase method, rather than on companies which did not amortise goodwill using pooling, or that were preferring to write off the bid premium in the first year as in-process research and development (Hopkins et al., 2000). As a result, the pooling ban eliminates the bias that analysts may have priced into the shares of pooled companies.

Perhaps the major drawback of the M&A accounting changes is the need of goodwill impairment charges, which is linked to the assumption that companies will fairly recognise situations where M&A deals may have not meet expectations. In this case, the new rules introduce an element of uncertainty

343 and additional risk into future earnings estimation due to the potential occurrence of impairment surprises. Nevertheless, stock prices are not likely to decline significantly for companies with one-time impairment write-offs unless they become habitual (Hopkins et al., 2000). This was indeed the case in early 2000’s, as companies profited immediately from a “big bath” earnings by the first time of the new FASB’s standards adoption (Jordan & Clark, 2004).

8.5 Conclusions

In this chapter, the impacts of the changes in business combinations accounting on M&A activity were explored further, in the context of a wider analysis of diverse evidence shown throughout this thesis. This broader analysis included not only the study of economic consequences from the accounting changes on M&A activity, but also the examination of impacts on financial reporting at international level.

If the impact of the accounting changes appears not to be materialised in managerial decisions on M&A deals, the same cannot be argued about managerial discretion in financial reporting. The accounting for business combinations and goodwill can affect significantly the financial reporting and provides opportunities for earnings management. Managerial discretion was a reality by the time when accounting choice was yet possible in business combinations, as shown by diverse literature documenting manipulation of M&A deals, often simply as a mean to comply for uniting of interests.

Unsurprisingly, FASB’s proposals generated great concern among several industries, leading to some fierce lobbying which resulted in a compromise from the standard-setting board, materialised by the replacement of goodwill and other intangibles amortisation by impairment tests. Consequently, it is suggested here that managerial discretion shifted to earnings management by means of impairment testing. Moreover, despite the limitations imposed by IASB on business combinations methods at international level, a number of

344 countries in Europe have a dual accounting system, allowing a significant number of companies to continue using pooling of interests. Further considerations regarding these issues are to be undertaken in the final chapter of this thesis which follows.

345

Chapter 9 Conclusions and Further Research

9.1 Introduction

From the examination made in the preceding chapters concerning the research hypotheses’ development and testing, it is suggested here that the changes in business combinations accounting did not produce significant effects on M&A activity in the USA. Nevertheless, evidence shown in chapter 5, and discussed in chapter 8, together with the early literature review made in this thesis, also shed light on some possibly significant collateral effects produced by FASB’s changes. Accordingly, the contextualisation of the central argument of this thesis has been enriched with the examination of the managerial perception of the impact of the new accounting rules on M&A activity, together with the discussion of the implications of managerial discretion on financial reporting under business combinations accounting, in the scope of the examination of the impacts of the new M&A accounting rules on financial reporting.

Despite the models developed for this thesis were carefully designed and closely checked, there are always limitations inherent to statistical modelling and testing. Moreover, the examination of additional evidence, in the scope of a wider analysis of the main research question, carries additional limitations, that will be outlined in the first section of this chapter.

The concluding chapter of this thesis also comprises suggestions for further research and generalisation of research findings. Finally, this chapter concludes

346 with considerations about the contributions of this thesis for the knowledge, particularly in what concerns to financial reporting related to business combinations accounting, and also to the existing literature in M&A activity.

9.2 Limitations

The development of the models used to test the research hypotheses benefited from several contributions mostly from literature in finance, economics, and accounting. The profuse existing literature on both business combinations accounting and M&A activity provided reasonably sound theoretical and methodological bases, allowing comprehensive models to develop in accordance with the research purposes.

The main methodology used in this thesis is based in multiple regression analysis. Regression models are approximations to reality, which is often complex enough to make it difficult to ensure the appropriateness of the modelling procedure. The M&A activity provides a fine example of a complex reality, as illustrated by the failed attempts to explain comprehensively the occurrence of M&A in waves. Nevertheless, this thesis aims a particular purpose, for which it is not necessary to capture the whole reality of M&A activity over more than a century. Following the examination of the pattern of M&A activity, by means of literature review and data analysis, there is confidence about the appropriateness of the modelling developed here. Moreover, as an in depth analysis was carried for the period subject to testing in this thesis, covering different time frames, and including several sensitivity analyses, it was ensured that the examination covered a multiplicity of different possible situations.

The inexistence of a comprehensive model that could be used straightforwardly as a benchmark, led to the development of concurrent models based in different contributions from literature, also enabling the testing of the research hypotheses under different dimensions. The use of concurrent models allowed

347 several of the testing results to undergo a cross-analysis, being expected that such confirmation procedure would have reinforced the reliability of the methodology used, strengthening the research findings obtained in this thesis. The use of concurrent models was also justified by the need to ensure a significant number of observations. On the one hand, being this research limited to a maximum time frame of analysis of 15 years, the models could not use a data aggregation superior to a quarter in order to guarantee a reasonable number of observations. On the other, the use of very-short periods of data aggregation, in some models, limited the number of exogenous explanatory variables possible to be utilised due to missing available data.

There is indeed a trade-off between the number of observations and the number of variables that can be used. For example, for a model using monthly data is relatively easy to find statistical data to prepare a reasonably large set of explanatory variables. Conversely, for very short-term studies, using weekly or daily data, the number of observations will be higher, but the number of variables possible to construct will be lower.

Therefore, the inexistence of a comprehensive theory, and some data aggregations used, may constitute limitations for the models developed in this thesis. Nevertheless, as discussed before, despite a wide-ranging theory and a complete model are missing, a generic regression model, appropriately designed and fitted, may work as a fine representation of reality, provided that partial theories and models are available in literature, which is the case, as proved by the many contributions to this thesis from diverse literature in M&A, econometrics, or in behavioural finance.

Besides the basic limitations intrinsic to regression analysis, there are other limitations related specifically with the procedure used. More precisely, related to the use of multiple regression analysis together with time series and stepwise backward elimination of variables. As an example of possible drawbacks, the

348 order of the variables entered in the regression model can influence the variables selection process in stepwise regression.273

Despite any possible adjustments and further fittings that could possibly result in an improvement of the research models, there is a conviction, however, that the versions used in this thesis assure appropriateness concerning the research hypothesis testing.274

Other limitations related to the models used in this thesis, include the small number of observations used in the models using monthly data, minimised by the fact that the data aggregation comes from sizeable samples; possible mistakes in M&A data provided by the data provider, Thomson Financial/Reuters, minimised by the use of large samples; and absence of a breakdown of M&A data by industry, which would allow performing a cross- -sectional analysis.

Besides the limitations tied to the models development and the testing of the research hypotheses, there are also limitations concerning additional evidence which has been examined, but that was not necessarily subject to statistical testing. It was the case of the examination of annual reports from S&P 500 firms. In the case of the survey results, the evidence collected has been tested, but using small samples. The justification for such procedures and the respective limitations were examined before in chapter 5.

273 In order to access for any possible procedure inconsistencies, and to be sure they were not meaningful, numerous sensitivity analyses were conducted – not necessarily shown in this thesis. For a list of drawbacks in stepwise regression see e.g. Stevens (2002). 274 As discussed before, this confidence is supported by the corroborative use of different sets of data, the exploration of appropriate fittings for the models, the analysis and treatment of outliers, the unconditional observation of the regression analysis assumptions, and by the numerous sensitivity analyses performed.

349

9.3 Summary of main research and generalisation

In this thesis, the controversial nature of business combinations accounting has been considered in the context of the financial reporting regulatory due process, together with the examination of the issues concerning the challenges raised by the existing user-orientated financial reporting, and by conflicting interests between different stakeholders, particularly whenever managerial discretion affects investors’ wealth. The history of accounting standard-setting policy provides several pieces of evidence revealing that GAAP tend to be set reactively rather than proactively, being the discussions on business combinations accounting a fine example of this. Alongside with some fierce corporate lobbying on the new proposals for business combinations accounting, high-level politics was also directly involved in the regulatory process. The examination of the interests underlying accounting choice helped to understand why the lobbying from some pressure groups was so intense: from banking and technological industries, to M&A financial advisors and analysts. Accordingly, the research hypotheses were developed in order to allow testing the appropriateness of FASB’s new business combinations GAAP, in the scope of the desired minimisation of the economic consequences caused by changes in accounting policy – as many argued that pooling of interests elimination would affect negatively the M&A activity and the economy of the USA.

The present research examines primarily the possible impacts on M&A activity, as a consequence of the adoption in 2001 of a new set of standards that changed dramatically the business combinations accounting in the USA. Two main time frames were examined: a larger set from 1994 to 2008, using quarterly data; and a shorter three-year period, from 2000 to 2002, using monthly, weekly, and daily data. As shown in chapter 7, in the sensitivity analysis section, the daily data aggregation set was also used for an examination of the very short-term, by testing several alternative time frames, from a few weeks to a few months time. As shown in chapter 6, the triennial M&A activity subject to analysis is represented by a sample of 20,664 M&A deals, comprising 19,758 completed deals and 906 withdrawals, while the larger sample of 1994-2008’s is composed

350 by a total of 127,888 M&A deals, including 122,871 completed deals and a total of 5,017 withdrawn deals.

A long list of explanatory variables was prepared, comprising different types of endogenous and exogenous variables to the M&A activity; and also key intervention variables, constructed with the purpose of testing the research hypotheses. Several models were drawn and fitted, supported by multiple regression analysis with time series methodology, and were accordingly subject to testing, using different sets of data aggregation. Stepwise backward elimination was used, not only to fit the models, but, more importantly, to enable assessing about the importance of the intervention variables. The use of stepwise regression resulted in the elimination of non-statistically significant variables at the 95% level.

Every testing was undertaken with the purpose of identifying potential impacts following the effectiveness of the new accounting standards, on 1 July 2001, by comparing the corresponding time periods adjacent to the event day. The adoption of the two main event periods of examination, 2000-2002, and 1994- -2008, has also allowed capturing any possible effects arising from the publication of FASB’s reviewed ED on business combinations accounting, on 14 February 2001 – a situation which would not be confirmed, as in any case the event variables were statistically significant, as to be discussed in the following paragraphs.

The event variables were not selected by any model constructed to test the research hypotheses. Several sensitivity analyses were also conducted, including the examination of possible effects from extreme observations, non-trading days and reduced trading days; alternative event time-frames; and alternative significance levels; with no significant implications on the main findings. Of particular importance, concerning the consistency of the regression methodology adopted, the analysis of possible outliers and influential points allowed concluding about the inexistence of significant abnormal effects on the regressions results, therefore avoiding the need for any adjustments, and

351 meaning that the research findings could be maintained as initially drawn. In resume, the results obtained from testing the research models proved to be sound enough allowing solid conclusions to be drawn.

As shown by the sensitivity analysis performed in chapter 7, the event variables were not even able to reach the 90% level of significance threshold, and therefore the research hypotheses, as outlined in chapter 4, could not be rejected. Therefore, the present research findings indicate that the effectiveness of the new accounting standards did not have any significant impact on the overall M&A activity in the USA, during the periods examined, covering from the very short-term to the long run. Exchange markets conditions, time, and other M&A endogenous factors, proved to be more influential than the changes in the accounting for business combinations in respect to the evolution of M&A activity.

The major generalisation that can be made from these findings is that changes in accounting rules do not necessarily produce economic consequences. GAAP changes are not neutral, and therefore its impacts should be carefully measured by the standard-setters. In the case of the new business combinations accounting standards, it seems that the compromise between the FASB and the lobbying opposing to the discontinuation of pooling of interests was successful.

Another possible generalisation is that the elimination of accounting choice in business combinations accounting did not result necessarily negative for the M&A activity and for the economy. Moreover, it has also probably improved the quality of the information provided to financial reporting users, by enhancing comparability and fairness in the M&A market, as purchased goodwill recognition was enforced for all companies involved in M&A deals.275

275 It seems unquestionable that pooling of interests elimination improved the quality of financial reporting. However the same cannot be argued straightforward for the replacement of purchased goodwill amortisation by impairment tests, as discussed before in chapters 5 and 8.

352

Finally, despite this thesis being focused on accounting changes occurred in the USA, its findings can be extrapolated worldwide, namely to the several existing national sets of accounting standards which continue to allow pooling of interests method to be used, as discussed in the final section of this thesis.

9.4 Suggestions for further research

A first suggestion for further research is to test the hypotheses formulated in this thesis using cross-sectional data, in order to examine the occurrence of possible impacts of the accounting changes at industry-level. As examined in earlier chapters, anecdotal evidence suggests that pooling of interests was largely favoured in some sectors of activity, such as IT, industrials, and financials. Furthermore, as shown in chapter 5, and discussed in chapter 8, there is some evidence collected in this thesis suggesting that M&A activity in IT industry could have suffered from some degree of constraint as a consequence of the adoption of the new M&A accounting rules. Although not statistically significant, evidence also suggests the existence of possibly meaningful impacts on financials industry. Such indications are worthwhile of further examination, by means of formal hypothesising and testing, as some light could be shed on how reasonable were the allegations, from several representatives of some industries, supporting that the changes in business combinations accounting could result adverse for the financial reporting of companies possibly interested to be involved in M&A deals, particularly for the ones preferring pooling of interests as a way to avoid recognising purchased goodwill and other intangible assets. It was therefore feared that FASB’s changes on M&A accounting could end as an inhibiting factor for a significant number of firms in some sectors of activity, negatively impacting the overall M&A activity, and resulting in possibly hazardous effects for the economy of the USA.

As mentioned before, research on pooling of interests versus purchase method may have come to an end, as the focus shifted to the issue of the impairment testing under SFAS 142. However, following adoption of SFAS 141 and SFAS

353

142, the “new” purchase method differs considerably from its previous version, primarily due to the replacement of constant goodwill amortisation charges by intermittent impairment losses. Therefore, it would be interesting to test capital market reactions to purchase method pre- and post-implementation of the new business combinations accounting standards.

Another suggestion for further research is to test comprehensively the impacts of the new M&A standards on reported earnings. As examined in chapter 5, the impact of the nonamortisation of purchased goodwill and certain intangible assets was very significant, and therefore deserves some further examination. Nevertheless, existing literature is only focused on the examination of impairment losses, disregarding the initial positive impact of such nonamortisation, and therefore neglecting a proper inclusive investigation of the effects resulting from the implementation of the new standards.

Finally, as this thesis examined the impact of the changes in M&A accounting on M&A activity in the USA, it would be utterly interesting if a similar research would be undertaken in other countries and regions worldwide. Since IFRS 3, in 2004, has also eliminated the pooling option, it is therefore possible to replicate the present research study in countries where IASB standards have been adopted. Regarding the examination of possible impacts on M&A activity and on financial reporting as a consequence of the changes occurred in business combinations accounting rules worldwide, the possibility of obtaining some international comparisons would be certainly of great interest.

9.5 Contributions

This thesis offers some breakthroughs in areas of research related to business combinations accounting and also to M&A activity. Indeed, the research developed here provides several advances in the knowledge of M&A activity and accounting, not only by means of original contributions, but also by means of corroborating, or questioning, several findings from existing literature.

354

To the best of the knowledge of the author of this thesis, the main research question formulated here is unique in the literature. As a consequence of the novelty presented by this thesis, the research hypotheses and the models developed for the due testing are also original in many ways. Similarly, the survey study and the examination of annual reports, do not find parallel in the literature, at least in terms of comprehensiveness.276

Therefore, this thesis represents a breakthrough in the research concerned with accounting choice in business combinations accounting in the post-SFAS 141 & 142 implementation period. Despite its original purposes, this thesis is, nevertheless, related to diverse literature in business combinations and in M&A activity.

In terms of general contributions for the knowledge, the findings shown and discussed in this thesis suggest that the compromise achieved between the FASB and the different pressure groups was overall successful.277 The initial reaction to FASB’s early proposals was of great concern, as many managers and other professionals in the M&A business feared that the accounting changes could constrain considerably the M&A activity. Lobbyists also argued and warned about the possible negative impact on the economy of the USA as a consequence of a limited level of M&A activity. In order to keep the pooling elimination proposal, the FASB had to compromise, replacing the initial proposal of amortising purchased goodwill over a maximum of twenty years - versus the former AICPA’s APB ceiling of forty years - by impairment tests. With the publication of SFAS 141 and SFAS 142, in 2001, the dual accounting

276 There are a number of papers presenting partial pieces of evidence about the prospective impacts of the new business combinations accounting rules on financial reporting, mostly published by the time when FASB’s proposals were being publicly discussed. There are also many testimonies available from managers about FASB’s proposals and deliberations on business combinations accounting, some of which were reproduced in this thesis. 277 Nevertheless, it is important to recall that there is evidence indicating some degree of impact on M&A activity at industry-level, particularly for IT industry, and also possibly for other sectors of activity, such as the case of financials and industrials.

355 for business combinations has finally come to an end in the USA. As suggested by the testing results obtained and discussed in this thesis, the M&A activity continued to flow barely impassive, driven primarily by financial, economic, time, and seasonal factors.

This research also contributes to the existing knowledge by presenting fresh evidence about significant impacts of the new business combinations accounting rules on financial reporting. Indeed, evidence for S&P 500 companies, concerning reported figures, revealed that had the new FASB’s pronouncements being implemented -1 or -2 years before the effective adoption date, and the diluted EPS would have increased on average 20 and 30%, respectively, being this finding unique in existing literature, to the best of the knowledge of the author of this thesis.

Furthermore, additional evidence shown in this research, also collected from annual reports of S&P 500 companies, supports the “big bath” earnings theory, as a considerable number of companies recognised very significant impairment losses immediately upon the first year of SFAS 142 adoption, a time of economic contraction (see e.g. Jordan & Clark, 2004, 2005).

In terms of key contributions of this thesis to the existing literature, the finding that the new FASB’s pronouncements did not result in significant effects on the overall M&A activity, both in the very-short, short, medium, and long terms, is consistent with pre-2001 literature findings, suggesting that the selection of a particular accounting method in business combinations, either pooling of interests or purchase accounting, is irrelevant, as markets can adjust to the differences in accounting methods (Chatraphorn, 2001: 75).This conclusion is also consistent with existing literature supporting market efficiency, which argues that the market is able to assess the underlying economics of a transaction, regardless the accounting procedure used (M. L. Davis, 1990; Hong et al., 1978; Lindenberg & Ross, 1999; Vincent, 1997). Indeed, several event studies examining companies that used accounting choice to adjust accounting figures found that such practices do not produce economic effects (Brealey &

356

Myers, 1996, T. Copeland & Weston, 1983, Hong et al., 1978). Therefore, this thesis corroborates the suggestion that investors ignore accounting numbers, such as earnings, in their investment decisions, namely regarding M&A valuation, since they do not consider them to have true economic significance.

Finally, the findings of this thesis have also implications for several existing national sets of accounting standards worldwide which continue to include accounting choice in business combinations accounting methods and/or in the accounting treatment for purchased goodwill and other intangible assets. Furthermore, many countries keep permitting the use of pooling of interests method and the amortisation of purchased goodwill, despite the concurrent use of IASB’s standards, which, in turn, has already abandoned such procedures in the meantime..

The accounting choice in business combinations ended in the USA in 2001, and the IASB would also soon later follow FASB’s steps, by issuing IFRS 3 in 2004. A move which can be considered as natural since both standard-setting boards had previously agreed in a convergence project on business combinations. In the scope of a long-term project, both FASB and IASB have inclusively issued recently revised versions of SFAS 141 and IFRS 3, more precisely on 4 December 2007 and 10 January 2008, respectively, in order to reinforce international convergence in business combinations accounting.

As discussed in chapter 8, the case of many countries in Europe is particularly interesting. Several European national sets of accounting standards continue permitting to use pooling of interests, or a similar method, and to amortise purchased goodwill. However, with the adoption of IASB’s standards by the EU in 2005, a dual accounting system was created, in which any listed company is compelled to prepare consolidated accounts according to IASB’s standards. Under IASB’s GAAP, i.e. IFRS 3 and related standards, pooling of interests and purchased goodwill amortisation are not allowed, similarly to SFAS 141 and SFAS 142. Therefore, listed companies with consolidated accounting cannot continue to use anymore pooling of interests, or a similar method, and have

357 necessarily to recognise purchased goodwill in their balance sheets. Likewise, firms reporting under IASB GAAP are now testing purchased goodwill for impairment losses, to the detriment of the amortisation procedure.

Despite the trends set by the FASB and the IASB, not followed by other notable boards such as the UK ASB though, for many companies using domestic GAAP sets nothing has changed in business combinations accounting, as the pooling option and the amortisations of goodwill continue to be a reality. Such context configures an imbalanced situation, in which a small number of companies that has to use IASB’s standards cannot qualify any M&A deal for pooling of interests anymore. This is in clear contrast to the majority of the remaining European companies which, in turn, can be regarded as being privileged since they can continue to avoid recognising purchased goodwill, benefiting from the exemption of having to subject their earnings to any possible charges as a result of amortisations or impairment procedures, provided that they are able to qualify the M&A deals as uniting of interests. As IASB’s standards are nowadays accepted by over one hundred countries, this incongruent situation is also a reality for several other countries worldwide.

It is therefore argued in this thesis that national accounting standard-setters, not only from Europe but also worldwide, should follow closely IASB’s GAAP on this subject matter, by eliminating the accounting choice in business combinations, and by consequently enforcing the recognition of goodwill whenever the deal value is superior to the total fair values of the assets and liabilities of the firm considered to be the M&A target. By discontinuing the pooling of interests option and by establishing a single standard rule for any company involved in a M&A deal worldwide, financial reporting comparability and fairness among companies would be dramatically improved, resulting in a very significant progress in the context of an increasingly global competition.

358

References

3M Company. (2002). SEC Filing: Form 10-K405 Pursuant to Section 13 or 15(d) of the Securities Exchange Act of 1934 - For the year ended December 31, 2001. Retrieved 2005 (several accessing days), http://www.edgar-online.com

3M Company. (2003). SEC Filing: Form 10-K Pursuant to Section 13 or 15(d) of the Securities Exchange Act of 1934 - For the year ended December 31, 2002. Retrieved 2005 (several accessing days), http://www.edgar- -online.com

Abdel-Khalik, A. R., Regier, P., & Reiter, S. A. (1989). Some thoughts on empirical research in positive theory. In T. J. Frecka (Ed.), The State of Accounting Research as We Enter the 1990’s (pp. 153–190). Champaign, IL: University of Illinois at Urbana-Champaign, Department of Accountancy.

Aboody, D., Kasznik, R., & Williams, M. (2000). Purchase versus pooling in stock-for-stock acquisitions: why do firms care? Journal of Accounting & Economics, 29(3), 261-286.

Accountancy Age. (2008). Fair value critics turn on guidance. Retrieved 2 February 2009, Accountancy Age, 14 October 2008, http://www.accountancyage.com

Accountancy Age. (2009). Investor group says FASB is compromised. Retrieved 5 July 2009, Accountancy Age, 23 June 2009, http://www.accountancyage.com

Accounting Standards Board. (1994a). Acquisitions and Mergers, Financial Reporting Standard No. 6 (Amended in 1997 and 1998). London, UK: Accounting Standards Board.

359

Accounting Standards Board. (1994b). Fair Values in Acquisition Accounting, Financial Reporting Standard No. 7. London, UK: Accounting Standards Board.

Accounting Standards Board. (1997). Goodwill and Intangible Assets, Financial Reporting Standard No. 10 (Amended in July and December 1998). London, UK: Accounting Standards Board.

Accounting Standards Board. (1998). Impairment of Fixed Assets and Goodwill, Financial Reporting Standard No. 11. London, UK: Accounting Standards Board.

Accounting Standards Board. Financial Reporting Council. (2005a). Business Combinations, Financial Reporting Exposure Draft (FRED) No. 36. London UK: Accounting Standards Board.

Accounting Standards Board. Financial Reporting Council. (2005b). Impairment of Assets (IAS 36), Financial Reporting Exposure Draft (FRED) No. 38. London UK: Accounting Standards Board.

Accounting Standards Board. Financial Reporting Council. (2005c). Intangible Assets (IAS 38), Financial Reporting Exposure Draft (FRED) No. 37. London UK: Accounting Standards Board.

Accounting Standards Board. Financial Reporting Council. (2009). Policy Proposal: The Future of UK GAAP, Consultation Paper. London UK: Accounting Standards Board.

Accounting Standards Steering Committee of the Institute of Chartered Accountants of England and Wales. (1975). The Corporate Report. London, UK: Institute of Chartered Accountants of England and Wales.

Adobe Systems Inc. (2002). SEC Filing: Form 10-K Pursuant to Section 13 or 15(d) of the Securities Exchange Act of 1934 - For the year ended November 30, 2001. Retrieved 2005 (several accessing days), http://www.edgar-online.com

360

Aktas, N., Bodt, E. d., & Roll, R. (2004). Market response to European regulation of business combinations. Journal of Financial and Quantitative Analysis, 39(4), 731-757.

Alexander, D., & Nobes, C. (2001). Financial Accounting: An International Introduction. Essex, UK: Financial Times / Prentice Hall. Pearson Education Limited.

American Accounting Association. (1966). A Statement of Basic Accounting Theory: American Accounting Association.

American Accounting Association. (1971). Report of the Committee on Accounting Theory Construction and Verification. The Accounting Review, 46(Supplement), 53-63.

American Accounting Association. (1978). Report of the Committee on The Social Consequences of Accounting Information: American Accounting Association Committee Reports.

American Accounting Association Committee on Concepts and Standards for External Financial Reports. (1977). Statement on Accounting Theory and Theory Acceptance: American Accounting Association.

American Accounting Association. Financial Accounting Standards Committee of Accounting Procedure. (1999). Response to FASB Invitation to Comment on Methods of Accounting for Business Combinations: Recommendations of the G4+1 for Achieving Convergence. Accounting Horizons, 13(3), 299-303.

American Bankers Association. (2009a). About the American Bankers Association. Retrieved 15 October 2009, http://www.aba.com/About+ABA/default.htm

American Bankers Association. (2009b). Government relations overview. Retrieved 15 October 2009, http://www.aba.com/Industry+Issues/GR_Overview.htm

361

American Institute of Accountants. (1934). Audits of Corporate Accounts, Reprinted in 1936 in George O. May, Twenty-five years of accounting responsibility, 1911-1936 (pp. 119-120). New York: American Institute of Accountants.

American Institute of Accountants; Committee on Accounting Procedure. (1944). Accounting for Intangible Assets, Accounting Research Bulletin No. 24. New York: American Institute of Accountants.

American Institute of Accountants; Committee on Accounting Procedure. (1950). Business Combinations, Accounting Research Bulletin No. 40. New York: American Institute of Accountants.

American Institute of Accountants; Committee on Accounting Procedure. (1953). Restatement and Revision of Accounting Research Bulletins, Accounting Research Bulletin No. 43. New York: American Institute of Accountants.

American Institute of Accountants; Committee on Accounting Procedure. (1957). Business Combinations, Accounting Research Bulletin No. 48. New York: American Institute of Accountants.

American Institute of Accountants; Committee on Terminology. (1953). Review and Resume, Accounting Terminology Bulletin No. 1. New York: American Institute of Accountants.

American Institute of Certified Public Accountants. (1968-2003). Accounting Trends and Techniques (Several ed.). New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants. (1977). Code of professional ethics: Concepts of professional ethics, rules of conduct, effective March 1, 1973, interpretations of rules of conduct, ethics rulings (1st ed.). New York: American Institute of Certified Public Accountants.

362

American Institute of Certified Public Accountants; (Trueblood Committee). (1973). Objectives of Financial Statements, Report of the Study Group on the Objectives of Financial Statements. New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; (Wheat Study Group). (1972). Establishing Financial Accounting Standards, Report of the Study on Establishment of Accounting Principles New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; Accounting Principles Board. (1962a). Accounting for the "Investment Credit", Accounting Principles Board Opinion No. 2. New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; Accounting Principles Board. (1962b). Statement by the Accounting Principles Board, Accounting Principles Board Statement No. 1. New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; Accounting Principles Board. (1964). Accounting for the "Investment Credit", Accounting Principles Board Opinion No. 4. New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; Accounting Principles Board. (1970a). Basic Concepts and Accounting Principles Underlying Financial Statements of Business Enterprises Accounting Principles Board Statement No. 4. New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; Accounting Principles Board. (1970b). Business Combinations, Accounting Principles Board Opinion No. 16. New York: American Institute of Certified Public Accountants.

363

American Institute of Certified Public Accountants; Accounting Principles Board. (1970c). Intangible Assets, Accounting Principles Board Opinion No. 17. New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; Accounting Principles Board. (1970d). Proposed APB Opinion, Business Combinations and Intangible Assets, Exposure Draft. New York: American Institute of Certified Public Accountants.

American Institute of Certified Public Accountants; Wheat Study Group. (1972). Establishing Financial Accounting Standards, Report of the Study on Establishment of Accounting Principles New York: American Institute of Certified Public Accountants.

Andersen, T., Bollerslev, T., Diebold, F., & Vega, C. (2007). Real-time price discovery in stock, bond and foreign exchange markets Journal of International Economics, 73(2), 251-277.

Anderson, J. C., & Louderback III, J. G. (1975). Income manipulation and purchase-pooling: some additional results. Journal of Accounting Research, 13(2), 338-343.

Andrade, G., Mitchell, M., & Stafford, E. (2001). New evidence and perspectives on mergers. Journal of Economic Perspectives, 15(2), 103-120.

Andrade, G., & Stafford, E. (2004). Investigating the economic role of mergers. Journal of Corporate Finance, 10(1), 1-36.

Anthony, R. N. (1963). Showdown on accounting principles. Harvard Business Review(May-June), 99-106.

Ariel, R. A. (1987). A monthly effect in stock returns. Journal of Financial Economics, 18(1), 161-174.

Armstrong, M. (1977). The politics of establishing accounting standards. Journal of Accountancy, 143(2), 76-79.

364

Armstrong, P. (1991). Contradiction and social dynamics in the capitalist agency relationship. Accounting, Organizations and Society, 16(1), 1-25.

Ashbaugh, H., LaFond, R., & Olsson, P. (2004). The value implications of accounting choice. Working paper.

Asquith, P. (1983). Merger bids, uncertainty, and stockholder returns. Journal of Financial Economics, 11(1-4), 51-83.

Association for Financial Professionals. (2000). Business Combinations Survey Report. Bethesda, MD: Association for Financial Professionals.

Association of American Railroads. (undated). Mergers: Efficiency is the Target, Background paper on mergers. Washington, DC: Association of American Railroads.

Auster, E. R., & Sirower, M. L. (2002). The dynamics of merger and acquisition waves: a three-stage conceptual framework with implications for practice. Journal of Applied Behavioral Science, 38(2), 216-244.

Ayal, I., & Hornik, J. (1986). Foreign source effects on response behavior in cross-national mail surveys. International Journal of Research in Marketing 3(3), 157-167.

Ayers, B. C., Lefanowicz, C. E., & Robinson, J. R. (2000). The financial statement effects of eliminating the pooling-of-interests method of acquisition accounting. Accounting Horizons, 14(1), 1-19.

Ayers, B. C., Lefanowicz, C. E., & Robinson, J. R. (2002). Do firms purchase the pooling method? Review of Accounting Studies, 7(1), 5-32.

Backman, J. (1970). Merger accounting: an economist's view. Financial Analysts Journal, 26(4), 39-48.

Ball, R. (2006). International Financial Reporting Standards (IFRS): pros and cons for investors. Accounting and Business Research, 36(Special Issue: International Accounting Policy Forum), 5-27.

365

Banerjee, A., & Eckard, E. W. (1998). Are mega-mergers anticompetitive? Evidence from the first great merger wave. The Rand Journal of Economics, 29(4), 803-827.

Banerjee, A., & Eckard, E. W. (2001). Why regulate insider trading? Evidence from the first great merger wave (1897-1903). American Economic Review, 91(5), 1329-1349.

Bank of America Business Capital. (2006). Critical Factors to Consider When Managing an M&A Deal Team, (Jeffrey A. Beuche) CapitalEyes: a Bank of America Business Capital e-newsletter on leveraged finance (May/June ed.).

Barkoulas, J. T., Baum, C. F., & Chakraborty, A. (2001). Waves and persistence in merger and acquisition activity. Economics Letters, 70(2), 237–243.

Barth, M. E., Landsman, W. R., & Lang, M. H. (2008). International accounting standards and accounting quality. Journal of Accounting Research, 46(3), 467-498.

Baskin, J. B., & Miranti Jr., P. J. (1997). A History of Corporate Finance. New York: Cambridge University Press.

Bates, D. M., & Watts, D. G. (1988). Nonlinear Regression Analysis and its Applications. New York: John Wiley & Sons, Inc.

Bayless, M., & Chaplinsky, S. (1996). Is there a window of opportunity for seasoned equity issuance? Journal of Finance, 51(1), 253-278.

Beattie, E. (2009). Bankers lobby against fair value: ABA wants G-20 to address accounting issues. Retrieved 30 September 2009, Accountancy Magazine, 14 September 2009, http://www.accountancymagazine.com

Beatty, A., Chamberlain, S., & Magliolo, J. (1996). An empirical analysis of the economic implications of fair value accounting for investment securities. Journal of Accounting and Economics, 22(1-3), 43-77.

366

Beaver, W. H. (1989). Financial Reporting: An Accounting Revolution. Englewood Cliffs, NJ: Prenctice Hall.

Beckenstein, A. R. (1979). Merger activity and merger theories: an empirical investigation. Antitrust Bulletin, 24(2), 105-128.

Becketti, S. (1986). Corporate mergers and the business cycle. Federal Reserve Bank of Kansas City Economic Review, 71(5), 13-26.

Belkaoui, A. (1985). Accounting Theory (2nd ed.). San Diego, CA: Harcourt, Brace, Jovanovich.

Belsley, D. A., Kuh, E., & Welsch, R. E. (1980). Regression Diagnostics: Identifying Influential Data and Sources of Collinearity. New York: John Wiley & Sons, Inc.

Bensa, D. A., Nagarb, V., Skinner, D. J., & Wong, M. F. (2003). Employee stock options, EPS dilution, and stock repurchases. Journal of Accounting and Economics, 36(1-3), 51-90.

Benston, G. (1978). The economic consequences of financial accounting statements. In FASB Economic Consequences of Financial Statements - Selected Papers (pp. 161-252). Norwalk, CT: Financial Accounting Standards Board.

Beresford, D. R. (1989). How well does the FASB consider the consequences of its work? Financial Executive, 5(2), 24-28.

Beresford, D. R. (2001). Congress looks at accounting for business combinations. Accounting Horizons, 15(1), 73-86.

Berry, M. A., Gallinger, G. W., & Henderson, G. V., Jr. (1990). Using daily stock returns in event studies and the choice of parametric test statistics. Quarterly Journal of Business Finance and Economics, 29(1), 70-85.

Berument, H., & Kiymaz, H. (2001). The day of the week effect on stock market volatility. Journal of Economics and Finance, 25(2), 181-193.

367

Beuche, J. A. (2006). Managing an M&A dream deal team, Denver Business Journal (March 31 ed.). Denver, CO: Denver Business Journal.

Bittlingmayer, G. (1985). Did antitrust policy cause the great merger wave? Journal of Law and Economics, 28(1), 77-118.

Black, B. S. (2000a). The first international merger wave (and the fifth and last US wave). University of Miami Law Review, 54(4), 799-818.

Black, B. S. (2000b). Is this the first international merger wave? M&A Lawyer, 4(4), 20-26.

Blair, M. M. (1989). An empirical test of the connection between free cash flow and takeover waves, mimeo: Brookings Institution.

Blair, M. M., Lane, S. J., & Schary, M. M. (1991). Patterns of Corporate Restructuring, 1955-1987. Working paper. Brookings Discussion Paper Number 91-1(January).

Blair, M. M., & Schary, M. A. (1993). Industry-level pressures to restructure. In M. M. Blair (Ed.), The Deal Decade: What Takeovers and Leveraged Buyouts Mean for Corporate Governance. Washington, DC: The Brookings Institution.

Blume, M. E., & Stambaugh, R. F. (1983). Biases in computed returns: an application to the size effect. Journal of Financial Economics, 12(3), 387- -404.

Box, G. P., & Jenkins, G. M. (1970). Time Series Analysis: Forecasting and Control. San Francisco: Holden-Day.

Brace, I. (1994). Questionnaire Design: How to Plan, Structure and Write Survey Material for Effective Market Research. London, UK: Kogan Page Limited.

Bradley, M., Desai, A., & Kim, E. H. (1983). The rationale behind interfirm tender offers. Journal of Financial Economics, 11(1-4), 183-206.

368

Branch, B., Jung, J., & Yang, T. (2001). The Monday merger effect. International Review of Financial Analysis, 10(1), 1-19.

Brealey, R., & Myers, S. (1984). Principles of Corporate Finance (2nd ed.). New York: McGraw-Hill, Inc.

Brealey, R., & Myers, S. (1996). Principles of Corporate Finance (5th ed.). New York: McGraw-Hill, Inc.

Brealey, R., Myers, S., & Allen, F. (2006). Principles of Corporate Finance (8th ed.). New York: McGraw-Hill, Inc.

Brenner, V. C., & Shuey, R. E. (1972). An empirical study of support for APB Opinion No. 16. Journal of Accounting Research, 10(1), 200-208.

Bricker, R. J., & Previts, G. J. (1990). The sociology of accountancy: a study of academic and practice community schisms. Accounting Horizons, 4(1), 1- -14.

Briloff, A. J. (1967). Dirty pooling. The Accounting Review, 42(3), 489-496.

Briloff, A. J. (1968). Distortions arising from pooling-of-interests accounting. Financial Analysts Journal, 24(2), 71-80.

Briner, R. F., & Fulkerson, C. L. (2001). Will elimination of pooling accounting reduce mergers and acquisitions? Multinational Business Review, 9(1), 9- -14.

Brockwell, P. J., & Davis, R. A. (1996). An Introduction to Time Series and Forecasting. New York: Springer-Verlag.

Bromwich, M. (1985). The Economics of Accounting Standard Setting. London, U.K.: Prentice Hall International / The Institute of Chartered Accountants in England and Wales.

Brooks, C. (2005). Introductory Econometrics for Finance. Cambridge, UK: Cambridge University Press.

369

Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: the case of event studies. Journal of Financial Economics, 14(1), 3-31.

Brown, V. (1990). Accounting standards: their economic and social consequences. Accounting Horizons, 4(3), 89-97.

Burchell, S., Clubb, C., Hopwood, A., Hughes, J., & Nahapiet, J. (1980). The roles of accounting organizations and society. Accounting, Organizations and Society, 5(1), 5-27.

Bureau of Economic Analysis. (2009). Mission, Vision, and Values. Retrieved 15 February 2009, http://www.bea.gov/about/mission.htm

Burns, J. B. (1998). Railroad Mergers and the Language of Unification. Westport, CT: Quorum Books.

Cabral, L. M. B. (2000). Introduction to Industrial Organization (Massachusetts Institute of Technology ed.). Cambridge, MA: MIT Press.

Cabral, L. M. B. (2005). An equilibrium approach to international merger policy. International Journal of Industrial Organization, 23(9-10), 739- -751.

Calomiris, C. W., & Karceski, J. (2000). Is the bank merger wave of the 1990s efficient? In S. N. Kaplan (Ed.), Mergers and Productivity (pp. 93-161). Chicago, IL: National Bureau of Economic Research.

Canning, J. B. (1929). The Economics of Accountancy. New York: Ronald Press.

Carmichael, D. R., Lilien, S. B., & Mellman, M. (1999). Accountants' Handbook (9th ed.). New York: John Wiley & Sons.

Cattlet, G., & Olson, N. (1968). Accounting for Goodwill, Accounting Research Study No. 10. New York: American Institute of Certified Public Accountants.

370

Chambers, R. J. (1963). Why bother with postulates? Journal of Accounting Research, 1(1), 3-15.

Chambers, R. J. (1979). Canning's the economics of accountancy: after 50 years. The Accounting Review, 54(4), 734-775.

Chand, P., & Cummings, L. (2008). The political and unstable nature of the iASB's 'stable platform': post-convergence Australian experience. Australian Accounting Review, 18(3), 175-184.

Chant, P. D. (1980). On the predictability of corporate earnings per share behavior. The Journal of Finance, 35(1), 13-21.

Chatraphorn, P. (2001). Accounting for Business Combinations: A Test for Long-Term Market Memory. Unpublished Doctor of Philosophy in Business Administration with a Concentration in Accounting, Virginia Polytechnic Institute and State University, Blacksburg, VA.

Choe, H., Masulis, R. W., & Nanda, V. (1993). Common stock offerings across the business cycle: theory and evidence. Journal of Empirical Finance, 1(1), 3-31.

Chowdhury, A. R. (1993). Univariate time-series behaviour of merger activity and its various components in the United States. Applied Financial Economics, 3(1), 61-66.

Chung, K. S., & Weston, J. F. (1982). Diversification and mergers in a strategic long-range-planning framework. In M. Keenan & L. J. White (Eds.), Mergers and Acquisitions: Current Problems in Perspective. Lexington, MA: D. C. Heath & Co.

Cisco Systems. (1998). SEC Filing: Form 10-K Pursuant to Section 13 or 15(d) of the Securities Exchange Act of 1934 - For the year ended July 25, 1998. Retrieved 2005 (several accessing days), http://www.sec.gov/edgar.shtml

371

Clark, J. J., Chakrabarti, A. K., & Chiang, T. C. (1988). Trend and stochastic movements in U.S. merger activity. Quarterly Review of Economics and Business, 28(2), 6-19.

Cole, W. M. (1908). Accounts, Their Construction and Interpretation for Business Man and Students of Affairs. Boston, MA: Houghton Mifflin Company.

Coleman, B. (1998). Daimler Chrsyler merger clears a hurdle as Daimler holders swap 97% of shares. The Wall Street Journal(102), A3.

Compustat database. (2006). New York: Standard and Poor’s Corporation.

Conn, R. L. (1973). Performance of conglomerate firms: comment. Journal of Finance, 28(3), 754-758.

Conn, R. L., Cosh, A., Guest, P. M., & Hughes, A. (2005). The impact on UK acquirers of domestic, cross-border, public and private acquisitions. Journal of Business Finance & Accounting, 32(5-6), 815-870.

Conrad, J., & Kaul, G. (1993). Long-term market overreaction or biases in computed returns? Journal of Finance, 48(1), 39-63.

Cooper, S., Crowther, D. E., Davies, M., & Davis, E. W. (2001). Shareholder or Stakeholder Value? The Development of Indicators for the Control and Measurement of Performance. London: CIMA.

Coopers & Lybrand. (1993). Making a Success of Acquisitions. London, UK: Coopers & Lybrand.

Copeland, R. M., & Wojdak, J. F. (1969). Income manipulation and the purchase-pooling choice. Journal of Accounting Research, 7(2), 188-195.

Copeland, T., Koller, T., & Murrin, J. (2000). Valuation: Measuring and Managing the Value of Companies (3rd ed.). New York: John Wiley & Sons.

372

Copeland, T., & Weston, J. F. (1983). Financial Theory and Corporate Policy. Reading, MA: Addison-Wesley Publishing Company.

Core, J., Guay, W., & Kothari, S. P. (2002). The economic dilution of employee stock options: diluted EPS for valuation and financial reporting. The Accounting Review, 77(3), 627-652.

Cornell, B. (1993). Corporate Valuation. Homewood, IL: Business One Irwin.

Covin, T. J., Kolenko, T. A., Sightler, K. W., & Tudor, R. K. (1997). Leadership style and post-merger satisfaction. Journal of Management Development, 16(1), 22-33.

Crowther, D. E. (2002a). Creating Shareholder Value. London: Spiro Press.

Crowther, D. E. (2002b). A Social Critique of Corporate Reporting: A Semiotic Analysis of Corporate Financial and Environmental Reporting. Aldershot: Ashgate.

CRSP database. (2006). Chicago: University of Chicago Center for Research on Securities Prices.

Curran, W. S. (1964). Book review: A critical study of accounting for business combinations - Wyatt, A. R. Journal of Finance, 19(1), 128-129.

Danos, P. (1977). A revolution in accounting thought? A comment. The Accounting Review, 52(3), 746-747.

Davidson III, W. N., Dutia, D., & Cheng, L. (1989). A re-examination of the market reaction to failed mergers. Journal of Finance, 44(4), 1077-1083.

Davidson III, W. N., Rosenstein, S., & Sundaram, S. (2002). An empirical analysis of cancelled mergers, board composition and ownership structure. Applied Financial Economics, 12(7), 485-491.

373

Davis, G. F., & Stout, S. K. (1992). Organization theory and the market for corporate control. A dynamic analysis of the characteristics of large takeover targets, 1980-1990. Administrative Science Quarterly, 37, 605- -633.

Davis, M. L. (1990). Differential market reaction to pooling and purchase methods. The Accounting Review, 65(3), 696-709.

Davis, M. L. (1991). APB 16: time to reconsider. Journal of Accountancy, 172(4), 99-107.

Davis, M. L. (1996). The purchase vs. pooling controversy: how the stock market responds to goodwill. Journal of Applied Corporate Finance, 9(1), 50-60.

Dawson, S., & Dickinson, D. (1988). Conducting international mail surveys: the effect of incentives on response rates with an industry population. Journal of International Business Studies, 19(3), 491-496. de Vaus, D. A. (2002). Surveys in Social Research (5th ed.). London, UK: Allen & Unwin and Routledge.

Deakin, E. B. (1989). Rational economic behavior and lobbying on accounting issues: evidence from the oil and gas industry. The Accounting Review, 64(1), 137-151.

Dechow, P., & Skinner, D. (2000). Earnings management: reconciling the views of accounting academics, practitioners, and regulators. Accounting Horizons, 14(2), 235-250.

Dechow, P., Sloan, R., & Sweeney, A. (1995). Detecting earnings management. The Accounting Review, 70(2), 193-225.

Dechow, P., Sloan, R., & Sweeney, A. (1996). Causes and consequences of earnings manipulation: an analysis of firms subject to enforcement actions by the SEC. Contemporary Accounting Research, 13(1), 1-36.

374

Deinzer, H. T. (1965). Development of Accounting Thought. New York: Holt, Rinehart & Winston.

DeLong, G. L. (2003). The announcement effects of U.S. versus non-U.S. bank mergers: do they differ? Journal of Financial Research, 26(4), 487-500.

Demski, J. S., Dopuch, N., Lev, B., Ronen, J., Searfoss, G., & Sunder, S. (1991). A Statement on the State of Academic Accounting, Statement to the Research Director of the American Accounting Association: American Accounting Association.

Derhy, A. (1999). À quelles logiques obéissent les vagues de fusions et acquisitions? Revue Française de Gestion, 122(1), 17-26.

Dickinson, A., & Peterson, D. R. (1995). Expectations of weekend and turn-of- -the-month mean return shifts implicit in index call option prices. Journal of Financial and Strategic Decisions, 8(3), 69-76.

Dicksee, L. R., & Tillyard, F. (1906). Goodwill and its Treatment in Accounts. London, UK: Reprinted in 1976 in “The History of Accounting Collection”, New York: Arno Press (3rd ed.).

Dieter, R. (1989). Is now the time to revisit accounting for business combinations. The CPA Journal, 59(7), 44-48.

Dillman, D. A. (1991). The design and administration of mail surveys. Annual Review of Sociology, 17(2), 225-249.

Dillman, D. A., Phelps, G., Tortora, R., Swift, K., Kohrell, J., Berck, J., et al. (2009). Response rate and measurement differences in mixed-mode surveys using mail, telephone, interactive voice response (IVR) and the Internet. Social Science Research, 38(1), 1-18.

Dodd, P. (1980). Merger proposals, management discretion, and stockholder wealth. Journal of Financial Economics, 8(2), 105-137.

375

Dong, M., Hirshleifer, D., Richardson, S., & Teoh, S. H. (2006). Does investor misvaluation drive the takeover market? Journal of Finance, 61(2), 725- -762.

Dougherty, C. (2002). Introduction to Econometrics (2nd ed.). Oxford, UK: Oxford University Press.

Draper, N. R., & Smith, H. (1981). Applied Regression Analysis (2nd ed.). New York, NY: John Wiley & Sons, Inc.

Duggal, R., & Cudd, M. (1998). Dual office holding, corporate governance and takeover gains. Managerial Finance, 24(2), 3-13.

Duncan, J. R., & Carleton, R. L. (1999). Will poolings survive? The CPA Journal, 69(1), 22-27.

Dunne, K. M. (1990). An empirical analysis of management's choice of accounting treatment for business combinations. Journal of Accounting & Economics, 9(2), 111-133.

Duvall, L., Jennings, R., Robinson, J., & II, R. B. T. (1992). Can investors unravel the effects of goodwill accounting? Accounting Horizons, 6(2), 1- -14.

Dymski, G. A. (2002). The global bank merger wave: implications for developing countries. Developing Economies, 40(4), 435-466.

Eigen, M. M. (1965). Is pooling really necessary? The Accounting Review, 40(3), 536-540.

Eis, C. (1969). The 1919-1930 merger movement in American industry. The Journal of Law and Economics, 12(2), 267-296.

Elbannan, M., & McKinley, W. (2006). A theory of the corporate decision to resist FASB standards: an organization theory perspective. Accounting, Organizations and Society, 31(7), 601-622.

376

Elgers, P. T., & Clark, J. J. (1980). Merger types and shareholder returns: additional evidence. Financial Management, 9(2), 66-72.

Ellert, J. C. (1976). Mergers, anti-trust law enforcement and stockholder returns. Journal of Finance, 31(2), 715-732.

Erickson, M., & Wang, S. (1999). Earnings management by acquiring firms in stock for stock mergers. Journal of Accounting & Economics, 27(2), 149- -176.

European Commission. (2000). Mergers and Acquisitions. European Economy, Supplement A(5/6).

European Commission. (2002). Regulation (EC) No. 1606/2002 of the European Parliament and of the Council of 19 July 2002 on the application of international accounting standards. Brussels, Belgium: Official Journal of the European Communities.

Evans, S. (2008). The history of the financial meltdown. Retrieved 30 October 2008, http://www.independent.co.uk/news/business/analysis-and- -features/the-history-of-the-financial-meltdown-936671.html

Fabozzi, F. J., Ferri, M. G., Fabozzi, T. D., & Tucker, J. (1988). A note on unsuccessful tender offers and stockholder returns. Journal of Finance, 43(5), 1275-1283.

Fama, E. F. (1970). Efficient capital markets: a review of theory and empirical work. Journal of Finance, 25(2), 383-417.

Fama, E. F. (1998). Market efficiency, long-term returns, and behavioral finance. Journal of Financial Economics, 49(3), 283-306.

Fama, E. F., Fisher, L., Jensen, M. C., & R. Roll. (1969). The adjustment of stock prices to new information. International Economic Review, 10(1), 1-21.

377

Fauli-Oller, R. (2000). Takeover waves. Journal of Economics & Management Strategy, 9(2), 189-210.

Fédération des Experts Comptables Européens (FEE). (2002). FEE Survey on Business Combinations. Brussels, Belgium.

Fields, T. D., Lys, T. Z., & Vincent, L. (2001). Empirical research on accounting choice. Journal of Accounting & Economics, 31(1-3), 255-307.

Fienberg, S. E. (2007). The Analysis of Cross-Classified Categorical Data (2nd ed.). New York, NY: Spring Science.

Financial Accounting Standards Board. (1976a). Conceptual Framework for Financial Accounting and Reporting: Elements of Financial Statements and Their Measurement, FASB Discussion Memorandum. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1976b). Scope and Implications of the Conceptual Framework Project. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1977). Financial Accounting and Reporting by Oil and Gas Producing Companies, Statement of Financial Accounting Standards No. 19. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1978). Objectives of Financial Reporting by Business Enterprises, Statement of Financial Accounting Concepts No. 1. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1979). Suspension of Certain Accounting Requirements for Oil and Gas Producing Companies, an amendment of FASB Statement No. 19, Statement of Financial Accounting Standards No. 25. Stamford, CT: Financial Accounting Standards Board.

378

Financial Accounting Standards Board. (1980a). Elements of Financial Statements of Business Enterprises, Statement of Financial Accounting Concepts No. 3. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1980b). Qualitative Characteristics of Accounting Information, Statement of Financial Accounting Concepts No. 2. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1982). Disclosures about Oil and Gas Producing Activities, an Amendment of FASB Statements 19, 25, 33, and 39, Statement of Financial Accounting Standards No. 69. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1984). Recognition and Measurement in Financial Statements of Business Enterprises, Statement of Financial Accounting Concepts No. 5. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1985a). Accounting for the Costs of Computer Software to Be Sold, Leased, or Otherwise Marketed, Statement of Financial Accounting Standards No. 86. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1985b). Elements of Financial Statements, Statement of Financial Accounting Concepts No. 6. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1985c). Employers' Accounting for Pensions, Statement of Financial Accounting Standards No. 87. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1993). Accounting for Certain Investments in Debt and Equity Securities, Statement of Financial Accounting Standards No. 115. Norwalk, CT: Financial Accounting Standards Board.

379

Financial Accounting Standards Board. (1997a). Earnings per Share, Statement of Financial Accounting Standards No. 128. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1997b). Issues Associated with the FASB Project on Business Combinations, Special Report. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1998). Methods of Accounting for Business Combinations: Recommendations of the G4+1 for Achieving Convergence, Invitation to Comment (Position Paper). Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1999a). Business Combinations and Intangible Assets, Exposure Draft. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (1999b). FASB Invitation to Comment on Methods of Accounting for Business Combinations: Recommendations of the G4+1 for Achieving Convergence, Analysis of Comment Letters. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2000). Using Cash Flow Information and Present Value in Accounting Measurements, Statement of Financial Accounting Concepts No. 7. Stamford, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2001a). Business Combinations, Statement of Financial Accounting Standards No. 141. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2001b). Business Combinations and Intangible Assets - Accounting for Goodwill, (Revised) Exposure Draft. Norwalk, CT: Financial Accounting Standards Board.

380

Financial Accounting Standards Board. (2001c). FASB Completes Business Combinations Project. Retrieved 5 July 2001, http://www.fasb.org

Financial Accounting Standards Board. (2001d). FASB Completes Business Combinations Project. Retrieved 30 July 2001, http://www.fasb.org

Financial Accounting Standards Board. (2001e). Goodwill and Other Intangible Assets, Statement of Financial Accounting Standards No. 142. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2006). Fair Value Measurements, Statement of Financial Accounting Standards No. 157. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2007a). Business Combinations, Statement of Financial Accounting Standards No. 141 (Revised 2007). Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2007b). Facts about FASB. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2007c). The Fair Value Option for Financial Assets and Financial Liabilities, Including an amendment of FASB Statement No. 115, Statement of Financial Accounting Standards No. 159. Norwalk, CT: Financial Accounting Standards Board.

Financial Accounting Standards Board. (2009). News Release 03/12/09: FASB Chairman Robert H. Herz Testifies on Mark-to-Market Accounting. Retrieved 30 April 2009, http://www.fasb.org

Fixler, D. J., & Grimm, B. T. (2008). The reliability of the GDP and GDI estimates. Survey of Current Business, 88(2), 16-32.

Flegm, E. H. (1984). Accounting. New York: Mc Graw-Hill.

Fligstein, N. (1990). The Transformation of Corporate Control. Cambridge, MA: Harvard University Press.

381

Fogarty, T. J., Hussein, M. A., & Ketz, J. E. (1994). Political aspects of financial accounting standard setting in the USA. Accounting, Auditing & Accountability Journal, 7(4), 24-46.

Franks, J. R., & Mayer, C. (1996). Hostile takeovers and the correction of managerial failure. Journal of Financial Economics, 40(1), 163-181.

Frary, R. B. (1996). Hints for designing effective questionnaires, Practical Assessment, Research & Evaluation, 5(3), Retrieved 28 February 2009, from http://PAREonline.net/getvn.asp?v=5&n=3

Fuller, K., Netter, J., & Stegemoller, M. (2002). What do returns to bidding firms tell us? Evidence from firms that make many acquisitions. Journal of Finance, 57(4), 1763-1793.

G4+1 (Organization), & International Accounting Standards Committee. (1998). G4+1 Position Paper: Recommendations for Achieving Convergence on the Methods of Accounting for Business Combinations, A Discussion Paper Issued for Comment by the Staff of the International Accounting Standards Committee. London, UK: International Accounting Standards Committee.

Gagnon, J. (1967). Purchase versus pooling of interests: the search for a predictor. Journal of Accounting Research, Supplement to Vol. 5(Empirical Research in Accounting: Selected Studies), 187-204.

Gagnon, J. (1971). The purchase-pooling choice: some empirical evidence. Journal of Accounting Research, 9(1), 52-72.

Galai, D., & Kedar-Levy, H. (2005). Day-of-the-week effect in high moments. Financial Markets, Institutions and Instruments, 14(3), 169-186.

Gaughan, P. A. (1996). Mergers, Acquisitions and Corporate Restructurings (1st ed.). New York: John Wiley & Sons.

Gaughan, P. A. (1999). Mergers, Acquisitions and Corporate Restructurings (2nd ed.). New York: John Wiley & Sons.

382

Gaughan, P. A. (2002). Mergers, Acquisitions and Corporate Restructurings (3th ed.). New York, NY: John Wiley & Sons.

Gaughan, P. A. (2007). Mergers, Acquisitions and Corporate Restructurings (4th ed.). New Jersey, NJ: John Wiley & Sons.

Gaughan, P. A., & Thomas Hodson Jr. (1993). Statistical Analysis in Wrongful Discharge/Title VII Litigation. In Gaughan and R.J. Thornton (Ed.), Litigation Economics. Greenwich, CT: JAI Press.

Gerboth, D. L. (1987). The conceptual framework: not definitions but professional values. Accounting Horizons, 1(3), 1-8.

Gibbons, M., & Hess, P. (1981). Day of the week effects and assets returns. Journal of Business, 54(4), 579-596.

Gibbons, R. D. (1994). Statistical Methods for Groundwater Monitoring. New York, NY: John Wiley & Sons, Inc.

Gil-Alana, L. A., & Peláez, R. F. (2008). The persistence of earnings per share Review of Quantitative Finance and Accounting, 31(4), 425-439.

Golbe, D. L., & White, L. J. (1988). A time-series analysis of mergers and acquisitions in the U.S. economy. In A. J. Auerbach (Ed.), Corporate Takeovers: Causes and Consequences (pp. 265-302). Chicago, IL: University of Chicago Press / NBER.

Golbe, D. L., & White, L. J. (1993). Catch a wave: the time series behavior of mergers. The Review of Economics and Statistics, 75(3), 493-499.

Goldberg, L. (1964). Book review: A critical study of accounting for business combinations - Wyatt, A. R. The Accounting Review, 39(1), 231-232.

Gordon, M. J. (1964). Postulates, principles and research in accounting. The Accounting Review, 39(2), 251-263.

383

Gordon, M. J., Horwitz, B. N., & Meyers, P. T. (1966). Accounting measurements and normal growth of the firm. In R. K. Jaedicke, Y. Ijiri & O. Nielsen (Eds.), Research in Accounting Measurement (pp. 221-231). Menasha, WI: American Accounting Association.

Gort, M. (1969). An economic disturbance theory of mergers. Quarterly Journal of Economics, 83(4), 624-642.

Gorton, D. E. (1991). The SEC decision not to support SFAS 19: a case study of the effect of lobbying on standard setting. Accounting Horizons, 5(1), 29-41.

Grady, P. (1965). Inventory of Generally Accepted Accounting Principles for Business Enterprises, Accounting Research Study No. 7. New York: American Institute of Certified Public Accountants.

Grimm, B. T. (2005). Alternative measures of U.S. economic activity in business cycles and business cycle dating. Working paper. Bureau of Economics, Federal Trade Commission, BEA Working Paper WP2005–05, 1-14.

Grinblatt, M., & Titman, S. (1998). Financial Markets and Corporate Strategy (International ed.): Irwin / McGraw-Hill.

Gu, A. Y. (2002). Information speed and stock market behavior. Review of the Academy of Finance, 2(1), 137–146.

Gu, A. Y., & Finnerty, J. (2002). The evolution of market efficiency: 103 years of the Dow. Review of Quantitative Finance and Accounting, 18(3), 219- -237.

Guerard, J. B. J. (1985). Mergers, stock prices, and industrial production: an empirical test of the Nelson hypothesis In O. D. Anderson (Ed.), Time Series Analysis: Theory and Practice (Vol. 7). Amsterdam, Netherlands: Elsevier.

Gujarati, D. N. (1995). Basic Econometrics (3rd ed.). New York: McGraw-Hill.

384

Hakansson, N. H. (1978). Where we are in accounting. A review of Statement on Accounting Theory and Theory Acceptance. The Accounting Review, 53(3), 717-725.

Halpern, P. J. (1973). Empirical estimates of the amount and distribution of gains to companies in mergers. Journal of Business, 46(4), 554-575.

Harford, J. (2005). What drives merger waves? Journal of Financial Economics, 77(3), 529-560.

Harzing, A.-W. (1997). Response rates in international mail surveys: results of a 22-country study International Business Review 6(6), 641-665.

Harzing, A.-W. (2006). Response styles in cross-national mail survey research: a 26-country study. The International Journal of Crosscultural Management, 6(2), 243-266.

Hatfield, H. R. (1909). Modern Accounting: Its Principles and Some of Its Problems. New York: D. Happleton and Company. Reissued in 1976 by Arno Press, New York.

Hatfield, H. R., Sanders, T. H., & Moore, U. (1938). A Statement of Accounting Principles. New York: American Institute of Accountants. Reprinted in 1959 by the American Accounting Association.

Haugen, R. A., & Jorion, P. (1996). The January effect: still there after all these years. Financial Analysts Journal, 52(1), 27-31.

Hayn, C., & Hughes, P. J. (2006). Leading indicators of goodwill impairment Journal of Accounting, Auditing & Finance, 21(3), 223-266.

Hazelkorn, T., Zenner, M., & Shivdasani, A. (2004). Creating value with mergers and acquisitions. Journal of Applied Corporate Finance, 16(2-3), 81-90.

385

Healy, P. M., & Wahlen, J. M. (1999). A review of the earnings management literature and its implications for standard setting. Accounting Horizons, 13(4), 365-383.

Hegel, G. F. W. (1822). The Philosophy of History. New York: Republished in 1956 by Dover Publications.

Hendriksen, E. S., & van Breda, M. F. (1992). Accounting Theory (5th ed.). Boston, MA: Irwin.

Hendry, D. F., & Mizon, G. E. (2005). Forecasting in the Presence of Structural Breaks and Policy Regime Shifts. In D. W. Andrews & J. H. Stock (Eds.), Identification and Inference for Econometric Models: Essays in Honor of Thomas Rothenberg (pp. 480-502). Cambridge, UK: Cambridge University Press.

Hinchman, G., 2000. (2000). The politics of accounting. Financial Executive, 16(6), 65-65.

Hines, R. D. (1988). Financial accounting: in communicating reality, we construct reality. Accounting, Organizations and Society, 13(3), 251-261.

Hirsch, J. A. (2004). Stock Trader's Almanac 2005 (38th ed.). Hoboken, NJ: John Wiley & Sons, Inc.

Hocking, R. R. (1976). The analysis and selection of variables in linear regression. Biometrics, 32(1), 1-49.

Holbrook, A., Krosnick, J., & Pfent, A. (2007). The Causes and Consequences of Response Rates in Surveys by the News Media and Government Contractor Survey Research Firms. In James M. Lepkowski, N. Clyde Tucker, J. Michael Brick, Edith D. De Leeuw, Lilli Japec, Paul J. Lavrakas, M. W. Link & R. L. Sangster (Eds.), Advances in Telephone Survey Methodology. New York, NY: John Wiley & Sons.

386

Holgate, P. A. (1990). The history and regulation of goodwill accounting. In P. M. (Ed.), Brand and Goodwill Accounting Strategies (pp. 9-19). Cambridge: Woodhead-Faulkner.

Holl, P., & Kyriazis, D. (1996). The determinants of outcome in UK takeover bids. International Journal of the Economics of Business, 3(2), 165-184.

Holmstrom, B., & Kaplan, S. N. (2001). Corporate governance and merger activity in the United States: making sense of the 1980s and 1990s. Journal of Economic Perspectives, 15(2), 121-144.

Holsen, R. C. (1963). Another look at business combinations. Journal of Accountancy, 116(1), 67-69.

Holthausen, R. W., & Leftwich, R. W. (1983). The economic consequences of accounting choice: Implications of costly contracting and monitoring. Journal of Accounting and Economics, 5(2), 77-117.

Hong, H., Kaplan, R. S., & Mandelker, G. (1978). Pooling vs. purchase: the effects of accounting for mergers on stock prices. The Accounting Review, 53(1), 31-47.

Hopkins, P. E., Houston, R. W., & Peters, M. F. (2000). Purchase, pooling, and equity analysts' valuation judgments. The Accounting Review, 75(3), 257- -281.

Horngren, C. T. (1973). The marketing of accounting standards. Journal of Accountancy, 137(4), 61-66.

Hubbard, R. G., & Palia, D. (1999). A reexamination of the conglomerate merger wave in the 1960s: an internal capital markets view. The Journal of Finance, 54(3), 1131-1152.

Hughes, A. (1993). Mergers and economic performance in the UK: a survey of empirical evidence 1950–1990. In Matthew Bishop & John Kay (Ed.), European Mergers and Merger Policy (pp. 9-95). Oxford: Oxford University Press.

387

Hunt, H. G., & Hogler, R. L. (1990). Agency theory as ideology: a comparative analysis based on critical legal theory and radical accounting. Accounting, Organizations and Society 15(5), 437-454.

Hunt, J., Lees, S., Grumbar, J. J., & Vivian, P. D. (1987). Acquisitions: The Human Factor. London, UK: London Business School.

Huson, M., Scott, T., & Wier, H. (2001). Earnings dilution and the explanatory power of earnings for returns. The Accounting Review, 76(4), 589-612.

Iglewicz, B., & Hoaglin, D. C. (1993). How to Detect and Handle Outliers (Vol. 16). Milwaukee, WI: American Society for Quality Control (ASQC) Quality Press.

Ijiri, Y. (1967). The Foundations of Accounting Measurement. Englewoods Cliffs, NJ: Prentice Hall.

Ijiri, Y. (1971). Critique of the APB fundamentals statement. Journal of Accountancy, 131(11), 43-50.

Ijiri, Y. (1975). Theory of Accounting Measurement, Studies in Accounting Research # 10. Sarasota, FL: American Accounting Association.

International Accounting Standards Board. (2004). Business Combinations, International Financial Reporting Standard No. 3. London, UK: International Accounting Standards Board.

International Accounting Standards Board. (2008). Business Combinations, International Financial Reporting Standard No. 3 (Revised 2007). London, UK: International Accounting Standards Board.

Investors Technical Advisory Committee. (2009). Proposals to Help Establish and Maintain FASB Independence, Unsolicited Comment Letter to the Financial Accounting Foundation. Norwalk, CT: Investors Technical Advisory Committee.

388

Jennings, R., Robinson, J., Thompson, R., & Duvall, L. (1996). The relation between accounting goodwill numbers and equity values. Journal of Business Finance and Accounting, 23(4), 513-533.

Jensen, M. C. (1993). The modern industrial revolution, exit, and the failure of internal control systems. Journal of Finance, 48(3), 831-880.

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305-360.

Jensen, M. C., & Ruback, R. (1983). The market for corporate control: the scientific evidence. Journal of Financial Economics, 11(1-4), 5-50.

Jobber, D., & Saunders, J. (1988). An experimental investigation into cross- -national mail survey response rates. Journal of International Business Studies, 19(3), 483-489.

Johnston, D., & Jones, D. A. (2006). How does accounting fit into a firm's political strategy? Journal of Accounting and Public Policy, 25(2), 195- -228.

Jordan, C. E., & Clark, S. J. (2004). Big bath earnings management: the case of goodwill impairment under SFAS No. 142. Journal of Applied Business Research, 20(2), 63-69.

Jordan, C. E., & Clark, S. J. (2005). Using goodwill impairment to accomplish big bath earnings management: an empirical study. Journal of Business and Economic Perspectives, 31(1), 80-87.

Jordan, C. E., Clark, S. J., & Vann, C. E. (2007). Using goodwill impairment to effect earnings management during SFAS No. 142’s year of adoption and later. Journal of Business & Economic Research, 5(1), 23-30.

Jovanovic, B., & Rousseau, P. (2002). The Q-theory of mergers. American Economic Review, papers and proceedings, 92(2), 198-204.

389

Judge, G. G., Hill, R. C., Griffiths, W. E., Lütkepohl, H., & Lee, T.-C. (1988). Introduction to the Theory and Practice of Econometrics (2nd ed.). New York: John Wiley & Sons, Inc.

Kam, V. (1990). Accounting Theory (2nd ed.). New York: John Wiley & Sons.

Kant, E. (1781). Critique of Pure Reason. London: Republished by Everyman in 1934.

Karamanou, I., & Nishiotis, G. P. (2009). Disclosure and the cost of capital: evidence from the market's reaction to firm voluntary adoption of IAS. Journal of Business Finance & Accounting, 36(7-8), 793-821.

Karpoff, J. M., Malatesta, P. H., & Walkling, R. A. (1996). Corporate governance and shareholder initiatives: empirical evidence. Journal of Financial Economics, 42(3), 365-395.

Katz, I., & Westbrook, J. (2009). Mark-to-market lobby buoys bank profits 20% as FASB may say yes. Retrieved 30 April 2009, Bloomberg, 30 March 2009, http://www.Bloomberg.com

Keef, S. P., & Roush, M. L. (2005). Day-of-the-week effects in the pre-holiday returns of the Standard & Poor's 500 stock index. Applied Financial Economics, 15(2), 107-119.

Keim, D. B. (1983). Size-related anomalies and stock return seasonality: further empirical evidence. Journal of Financial Economics, 12(1), 13-32.

Kelly-Newton, L. (1980). Accounting Policy Formulation: The Role of Corporate Management. Reading, MA: Addison-Wesley Publishing Company.

Kelly, L. (1985). Corporate management lobbying on FAS No. 8: some further evidence. Journal of Accounting Research, 23(2), 619-632.

390

Kennedy, V. A., & Limmack, R. J. (1996). Takeover activity, CEO turnover, and the market for corporate control. Journal of Business Finance and Accounting, 23(2), 267-285.

Ketz, E. (2003). Hidden Financial Risk: Understanding Off-Balance Sheet Accounting. Hoboken, NJ: John Wiley & Sons.

King, A. M. (2000). Silicon Valley defends 'pooling of interests'. Strategic Finance, 81(10), 76-76.

King, A. M., & Kelly, N. (2000). Merger accounting magic may disappear. Strategic Finance, 81(1), 39-43.

Kirk, D. (1978). Business and the FASB: the need for effective interaction. Management Accounting, 60(3), 17-20.

Kirk, D. (1986). Can you hear me now? CPA Journal, 56(67-70).

Kuhn, T. S. (1962). The Structure of Scientific Revolution. Chicago, IL: University of Chicago Press.

Kummer, C., & Steger, U. (2008). Why merger and acquisition (M&A) waves reoccur: the vicious circle from pressure to failure. Strategic Management Review, 2(1), 44-63.

Lakonishok, J., & Smidt, S. (1988). Are seasonal anomalies real? A ninety-year perspective. The Review of Financial Studies, 1(4), 403-425.

Lambrecht, B. M. (2004). The timing and terms of mergers motivated by economies of scale. Journal of Financial Economics, 72(1), 41-62.

Lamoreaux, N. R. (1985). The Great Merger Movement in American Business, 1895-1904 New York: Cambridge University Press.

Langetieg, T. C. (1978). An application of a three-factor performance index to measure stockholder gains from merger. Journal of Financial Economics, 6(4), 365-383.

391

Larcker, D. F. (2003). Discussion of employee stock options, EPS dilution, and stock repurchases' Journal of Accounting & Economics, 36(1-3), 45-49.

Laughlin, R. C. (1987). Accounting systems in organisational contexts: a case for critical theory. Accounting, Organizations & Society, 12(5), 479-502.

Laux, C., & Leuz, C. (2009). The crisis of fair-value accounting: making sense of the recent debate. Accounting, Organizations and Society, 34(6-7), 826- -834.

Leeth, J. D., & Borg, J. R. (2000). The impact of takeovers on shareholder wealth during the 1920s merger wave. Journal of Financial and Quantitative Analysis, 35(2), 217-238.

Leftwich, R. (1981). Evidence on the impact of mandatory accounting changes in accounting principles on corporate loan agreements. Journal of Accounting & Economics, 3(1), 3-36.

Leslie, L. L. (1972). Are high response rates essential to valid surveys? Social Science Research, 1(3), 323-334.

Lev, B. (2003). Corporate earnings: facts and fiction. Journal of Economic Perspectives, 17(2), 27-50.

Lev, B., & Mandelker, G. (1972). The microeconomic consequences of corporate mergers. Journal of Business, 45(1), 85-104.

Lewis, E. E., Lippitt, J. W., & Mastracchio, N. J. (2001). Users' comments on SFAS 141 and 142 on business combinations and goodwill. The CPA Journal, 71(10), 27-30.

Lindenberg, E., & Ross, M. P. (1999). To purchase or to pool? Does it matter? Journal of Applied Corporate Finance, 12(2), 32-47.

Linn, S. C., & Zhu, Z. (1997). Aggregate merger activity: new evidence on the wave hypothesis. Southern Economic Journal, 64(1), 130-146.

392

Linsmeier, T. J., Boatsman, J. R., Herz, R. H., Jennings, R. G., Jonas, G. J., Lang, M. H., et al. (1998). Response to FASB Special Report, “Issues Associated with the FASB Project on Business Combinations". Accounting Horizons, 12(1), 87-89.

Lintner, J. (1971). Expectations, mergers and equilibrium in purely competitive securities markets. American Economic Review, 61(2), 101-111.

Lomax, R. G. (2007). An Introduction to Statistical Concepts (2th ed.). Mahwah, NJ: Lawrence Erlbaum Associates.

Louis, H. (2004). Earnings management and the market performance of acquiring firms. Journal of Financial Economics, 74(1), 121-148.

Lowe, A., & Locke, J. (2005). Perceptions of journal quality and research paradigm: results of a web-based survey of British accounting academics. Accounting, Organizations and Society, 30(1), 81-98.

Lowe, A., Puxty, A. G., & Laughlin, R. C. (1983). Simple theories for complex processes: accounting policy and the market for myopia. Journal of Accounting and Public Policy, 2(1), 19-42.

Lowry, M., & Schwert, G. W. (2002). IPO market cycles: bubbles or sequential learning? Journal of Finance, 57(3), 1171-1200.

Lubatkin, M. (1983). Mergers and the performance of the acquiring firm. Academy of Management Review, 8(2), 218-225.

Lucas, D. J., & McDonald, R. L. (1990). Equity issues and stock price dynamics. Journal of Finance, 45(4), 1019-1043.

Lukka, K., & Kasanen, E. (1996). Is accounting a global or a local discipline? Evidence from major research journals. Accounting, Organizations and Society, 21(7), 755-773.

393

Lynch, H. H. (1971). Financial Performances of Conglomerates. Boston, MA: Division of Research, Graduate School of Business Administration, Harvard University.

Lys, T., & Vincent, L. (1995). An analysis of value destruction in AT&T’s acquisition of NCR. Journal of Financial Economics, 39(2-3), 353-378.

Maines, L. A., Bartov, E., Fairfield, P. M., Hirst, D. E., Iannaconi, T. A., Mallett, R., et al. (2003). Implications of accounting research for the FASB's initiatives on disclosure of information about intangible assets. Accounting Horizons, 17(2), 175-185.

Makridakis, S., Wheelwright, S. C., & Hyndman, R. J. (1998). Forecasting: Methods and Applications (3th ed.). New York: John Wiley & Sons, Inc.

Maksimovic, V., & Philips, G. (2001). The market for corporate assets: who engages in mergers and asset sales and are there efficiency gains? Journal of Finance, 56(6), 2019-2065.

Mandelker, G. (1974). Risk and return: the case of merging firms. Journal of Financial Economics, 1(4), 303-335.

Mansfield, E. (1986). Basic Statistics: With Applications. New York, NY: W. W. Norton & Company Ltd.

Marais, M. L. (1984). An application of the bootstrap method to the analysis of squared, standardized market model prediction errors. Journal of Accounting Research, 22(Supplement), 34-54.

Markham, J. W. (1955). Survey of the evidence and findings on mergers. In Business concentration and price policy (pp. 141-212). Princeton, NJ: Princeton University Press.

Markham, J. W. (1973). Conglomerate Enterprise and Public Policy. Boston, MA: Division of Research, Graduate School of Business Administration, Harvard University.

394

Markowitz, H. M. (1952). Portfolio selection. Journal of Finance, 7(1), 77-91.

Marquardt, C. A., & Wiedman, C. I. (2005). Earnings management through transaction structuring: contingent convertible debt and diluted earnings per share Journal of Accounting Research, 43(2), 205-243.

Martinez-Jerez, F. (2001). Capital markets assessment of accounting for business combinations. Harvard University Working paper.

Martinez-Jerez, F. (2003). Interaction between accounting and corporate governance: evidence from business combinations. Harvard University Working paper.

Massoud, M., & Raiborn, C. (2003). Accounting for goodwill: are we better off? Review of Business, 24(2), 26-32.

Matsusaka, J. G. (1993). Takeover motives during the conglomerate merger wave. The Rand Journal of Economics, 24(3), 357-379.

Matsushima, N. (2001). Horizontal mergers and merger waves in a location model. Australian Economic Papers, 40(3), 263-286.

Mattessich, R. (1980). On the evolution of theory construction in accounting: a personal account. Accounting and Business Research, 10(Special Accounting History Issue), 158-173.

Mattessich, R. (1995). Critique of Accounting: Examination of the Foundations and Normative Structure of an Applied Discipline. Westport, CT: Quorum Books.

Mattli, W., & Büthe, T. (2005). Accountability in accounting? The politics of private rule-making in the public interest. Governance: An International Journal of Policy, Administration, and Institutions, 18(3), 399-429.

McNown, R. (1986). On the uses of econometric models: a guide for policy makers. Policy Sciences, 19(4), 359-380.

395

Mead, W. J. (1969). Instantaneous merger profit as a conglomerate merger motive. Western Economic Journal, 7(4), 295-306.

Melicher, R. W., Ledolter, J., & D'Antonio, L. J. (1983). A time series analysis of aggregate merger activity. The Review of Economic Statistics, 65(3), 423-430.

Mergerstat. (2001). Mergerstat Review 2001. Los Angeles, California: FactSet Mergerstat, LLC.

Mergerstat. (2002). Mergerstat Review 2002. Los Angeles, California: FactSet Mergerstat, LLC.

Mergerstat. (2003). Mergerstat Review 2003. Los Angeles, California: FactSet Mergerstat, LLC.

Mezias, S., & Chung, S. (1989). The Process and Participation at FASB: A Study of the Comment Period. Florham Park, NJ: Financial Executives Research Foundation.

Miller, P. B. W. (1990). Should the FASB be neutral or responsive? Journal of Accountancy, 16(3), 35-38.

Miller, P. W., Redding, R. J., & Bahnson, P. R. (1998). The FASB: The People, the Process, and the Politics (Vol. 4th). Boston, MA: Irwin McGraw Hill.

Mitchell, M. L., & Mulherin, J. H. (1996). The impact of industry shocks on takeover and restructuring activity. Journal of Financial Economics, 41(2), 193-229.

Mitchell, M. L., Pulvino, T., & Stafford, E. (2004). Price pressure around mergers. The Journal of Finance, 59(1), 31-63.

Mitchell, M. L., & Stafford, E. (2000). Managerial decisions and long-term stock price performance. Journal of Business, 73(3), 287-320.

396

Mizon, G. E. (2004). The Use of Econometric Models in Economic Policy Analysis In A. Welfe (Ed.), New Directions in Macromodelling (Vol. 269, pp. 147-171). Amsterdam, Netherlands: Elsevier.

Moehrle, S. R., Reynolds-Moehrle, J. A., & Wallace, J. S. (2001). Pooling and rescinded or foregone stock repurchases. Accounting and the Public Interest, 1(1), 115-133.

Moeller, S., & Brady, C. (2007). Intelligent M&A: Navigating the Mergers and Acquisitions Minefield. Chichester, UK: John Wiley & Sons, Ltd.

Moeller, S. B., Schlingemann, F. P., & Stulz, R. M. (2005). Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave. Journal of Finance, 60(2), 757-782.

Montgomery, D. C., Peck, E. A., & Vining, G. G. (2001). Introduction to Linear Regression Analysis (3rd ed.). New York: John Wiley & Sons, Inc.

Montgomery, R. H. (1912). Auditing Theory and Practice. New York: The Ronald Press Company. Reprinted in 1927 by the The Ronald Press Company, New York (4th ed.).

Moonitz, M. (1961). The Basic Postulates of Accounting, Accounting Research Study No. 1. New York: American Institute of Certified Public Accountants.

Moonitz, M. (1974). Obtaining Agreement on Standards in the Accounting Profession, Studies in Accounting Research # 8. Sarasota, FL: American Accounting Association.

Moonitz, M., & Sprouse, R. T. (1962). A Tentative Set of Broad Accounting Principles for Business Enterprises, Accounting Research Study No. 3. New York: American Institute of Certified Public Accountants.

397

Morgan, G. (1988). Accounting as reality construction: towards a new epistemology for accounting practice. Accounting, Organizations & Society, 13(5), 477-483.

Mosich, A. N. (1968). Retroactive poolings in corporate mergers. Journal of Business, 41(3), 352-362.

Most, K. S., & Winters, A. L. (1977). Focus on standard setting: from Trueblood to the FASB. Journal of Accountancy, 143(2), 67-75.

Mouck, T. (1992). The rhetoric of science and the rhetoric of revolt in the ‘story’ of positive accounting theory. Accounting Auditing & Accountability Journal, 5(4), 35-56.

Mouck, T. (1993). The ’revolution‘‘ in financial reporting theory: a Kuhnian interpretation. The Accounting Historians Journal, 20(1), 33-57.

Muehlfeld, K., Sahib, P. R., & van Witteloostuijn, A. (2007). Completion or abandonment of mergers and acquisitions: evidence from the newspaper industry, 1981-2000. Journal of Media Economics, 20(2), 107-137.

Mueller, D. C. (1969). A theory of conglomerate mergers. Quarterly Journal of Economics, 83(4), 643-659.

Mueller, D. C. (1980). The Determinants and Effects of Mergers. Cambridge, MA: Oelgeschlager, Gunn & Hain Publishers, Inc.

Mueller, D. C. (1989). Mergers: causes, effects and policies. International Journal of Industrial Organization, 7(special issue on mergers), 1-10.

Mueller, D. C. (1997). Merger policy in the United States: a reconsideration. Review of Industrial Organization, 12(5–6), 655-685.

Myers, R. H. (1990). Classical and Modern Regression with Applications (2nd ed.). Boston, MA: Duxbury Press.

Nathan, K. (1988). Do firms pay to pool: some empirical evidence. Journal of Accounting and Public Policy, 7(3), 185-200.

398

Nathan, K., & Dunne, K. M. (1991). The purchase-pooling choice - some explanatory variables. Journal of Accounting and Public Policy, 10(4), 309-323.

National Bureau of Economic Research. (2007). US Business Cycle Expansions and Contractions. Retrieved 3 January 2007, http://www.nber.org/cycles

National Bureau of Economic Research. (2009). The NBER's business cycle dating procedure: frequently asked questions. Retrieved 14 February 2009, http://www.nber.org/cycles/recessions_faq.html

Nelson, K., & Strawser, R. H. (1970). A note on APB Opinion No. 16. Journal of Accounting Research, 8(2), 284-289.

Nelson, R. H. (1953). The momentum theory of goodwill. The Accounting Review, 28(4), 491-499.

Nelson, R. L. (1959). Merger Movements in American Industry: 1895-1956. Princeton, NJ: Princeton University Press.

Nelson, R. L. (1966). Business cycle factors in the choice between internal and external growth. In W. W. Alberts & J. E. Segall (Eds.), The Corporate Merger (pp. 52-66). Chicago: University of Chicago Press.

New York Stock Exchange. (2006a). History of New York Stock Exchange Holidays. Retrieved 2 January 2006, http://www.nyse.com/pdfs/closings.pdf

New York Stock Exchange. (2006b). New York Stock Exchange Special Closings, 1885-date. Retrieved 2 January 2006, http://www.nyse.com/pdfs/closings.pdf

Nitzan, J. (2001). Regimes of differential accumulation: mergers, stagflation and the logic of globalization. Review of International Political Economy, 8(2), 226-274.

399

Nobes, C. (1992). A political history of goodwill in the UK: an illustration of cyclical standard setting. Abacus, 28(2), 142-167.

Nobes, C., & Parker, R. (2002). Comparative International Accounting (7th ed.). Essex, UK: Financial Times / Prentice Hall. Pearson Education Limited.

Norris, K. B., & Ayres, F. L. (2000). Security market reaction to purchase business combinations at the first earnings announcement date. Journal of Business Research, 49(1), 79-88.

Ott, R. L. (1992). An Introduction to Statistical Methods and Data Analysis (4th ed.). Belmont, CA: Duxbury Press.

Palepu, K. G., Healy, P. M., & Bernard, V. L. (2000). Business Analysis and Valuation. Cincinnati, OH: SouthWestern College Publishing.

Paton, W. A. (1922). Accounting Theory - With Special Reference to the Corporate Enterprise. New York: The Ronald Press Company. Reprinted in 1973 by Scholars Book Company, Houston, TX.

Paton, W. A. (1936). A tentative statement of accounting principles underlying corporate financial statements. The Accounting Review(3), 187-191.

Paton, W. A. (1943). Accountants' Handbook (3th ed.). New York: John Wiley & Sons.

Paton, W. A., & Littleton, A. C. (1940). An Introduction to Corporate Accounting Standards, American Accounting Association Monograph No. 3: American Accounting Association.

Pautler, P. A. (2001). Evidence on mergers and acquisitions. Working paper. Bureau of Economics, Federal Trade Commission, 1-84.

Penman, M. E. (1970). What net asset value? An extension of a familiar debate. The Accounting Review, 45(2), 333-346.

400

Pfeffer, J., & Salancik, G. R. (2003). The External Control of Organizations: A Resource Dependence Perspective. Stanford, CA: Stanford University Press.

Phillips, G. M. (1995). Introduction to cross tabulation statistics for disparate impact cases. Litigation Economics Digest, 1(1), 39-44.

Phillips, L. C. (1965). Accounting for business combinations. The Accounting Review, 40(2), 377-381.

Pickering, J. F. (1978). The abandonment of major mergers in the U.K., 1965- -1975. Journal of Industrial Economics, 27(2), 123-131.

Piette, M. J. (1992). Methodological issues when using simple models to investigate employment discrimination: reply. Journal of Forensic Economics, 6(1), 43-50.

Popper, K. (1963). Conjectures and Refutations. London, UK: Routledge & Kegan Paul.

Preinreich, G. A. D. (1939). Economic theories of goodwill. Journal of Accountancy, 68(9), 169-180.

Prepared Testimony of Mr. Dennis Powell Vice President and Corporate Controller Cisco Systems, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

Prepared Testimony of Mr. Harvey Golub Chairman and CEO American Express, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

Prepared Testimony of Mr. James Barksdale Partner The Barksdale Group, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

Prepared Testimony of Mr. John Doerr Partner Kleiner, Perkins, Caufield & Byers, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

401

Prepared Testimony of Mr. Martin Regalia Chief Economist U.S. Chamber of Commerce, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

Prepared Testimony of Mr. Robert Ryan Chief Financial Officer Medtronic, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

Pryor, F. L. (2001). Dimensions of the worldwide merger boom. Journal of Economic Issues, 35(4), 825-840.

Quattrone, P. (2000). Constructivism and accounting research: towards a trans- -disciplinary perspective. Journal: Accounting, Auditing & Accountability Journal, 13(2), 130-155.

Radebaugh, L., & Gray, S. (1997). International Accounting and Multinational Enterprises (4th ed.). New York: John Wiley & Sons.

Radebaugh, L. H., Gebhardt, G., & Gray, S. J. (1995). Foreign stock exchange listings: a case study of Daimler-Benz Journal of International Financial Management and Accounting, 6(2), 158-192.

Ramanna, K. (2008). The implications of unverifiable fair-value accounting: evidence from the political economy of goodwill accounting. Journal of Accounting and Economics, 45(2-3), 253-281.

Ramanna, K., & Roychowdhury, S. (2008). Accounting information as political currency. Working paper 08-100.

Rappaport, A. (1977). The economic impact of financial accounting standards - implications for the FASB. In 1976 Accounting Research Convocation. Emerging Issues (pp. 113-128). AL: University of Alabama.

Rappaport, A. (1986). Creating Shareholder Value. New York: The Free Press.

Rau, P. R., & Stouraitis, A. (2006). Corporate event waves. Working Paper.

402

Ravenscraft, D. J. (1987). The 1980’s merger wave: an industrial organization perspective. In L. E. Brown & E. S. Ronsengren (Ed.), The Merger Boom (pp. 17-37). New Hampshire, MA: Federal Reserve Bank of Boston.

Rayburn, F. R., & Powers, O. S. (1991). A history of pooling of interests accounting for business combinations in the United States. The Accounting Historians Journal, 18(2), 155-192.

Reed, S. F., Lajoux, A. R., & Nesvold, H. P. (2007). The Art of M&A: A Merger Acquisition Buyout Guide (4th ed.). New York, NY: McGraw- -Hill.

Reid, S. R. (1968). Mergers, Managers and the Economy. New York: McGraw- -Hill Book Co.

Reid, S. R. (1975). The microeconomic consequences of corporate mergers: comment. Journal of Business, 48(2), 275-280.

Reiter, S. A., & Williams, P. F. (2002). The structure and progressivity of accounting research: the crisis in the academy revisited. Accounting, Organizations and Society, 27(6), 575-607.

Resende, M. (1999). Wave behaviour of mergers and acquisitions in the UK: a sectoral study. Oxford Bulletin of Economics and Statistics, 61(1), 85-94.

Rhodes-Kropf, M., Robinson, D. T., & Viswanathan, S. (2005). Valuation waves and merger activity: the empirical evidence. Journal of Financial Economics, 77(3), 561-603.

Rhodes-Kropf, M., & Viswanathan, S. (2004). Market valuation and merger waves. The Journal of Finance, 59(6), 2685-2718.

Ribeiro, H. R., & Crowther, D. E. (2005). Accounting for business combinations: (un)desirable uniformity? Paper presented at the British Accounting Association Annual Conference, Edinburgh, United Kingdom.

403

Ribeiro, H. R., & Crowther, D. E. (2006). Reaction to new Mergers & Acquisitions accounting rules and its impacts in the U.S.A. Paper presented at the British Accounting Association Annual Conference, Portsmouth, United Kingdom.

Ribeiro, H. R., & Crowther, D. E. (2007a). Did the accounting for business combinations produce any effects on M&A activity? Paper presented at the British Accounting Association Annual Conference, London, United Kingdom.

Ribeiro, H. R., & Crowther, D. E. (2007b). Impact of the FASB’s business combinations pronouncements on M&A activity. Paper presented at the XXXth European Accounting Association Annual Congress, Lisbon, Portugal.

Ribeiro, H. R., & Crowther, D. E. (2007c). The new accounting era of the purchase method and the impairment of purchased goodwill: what did really change in the U.S.A.? Paper presented at the XXXth European Accounting Association Annual Congress, Lisbon, Portugal.

Ribeiro, H. R., & Crowther, D. E. (2008a). Accounting for business combinations: lessons from the U.S.A.? Paper presented at the XXXIst European Accounting Association Annual Congress, Rotterdam, Netherlands.

Ribeiro, H. R., & Crowther, D. E. (2008b). Business combinations accounting in the United States from AICPA to FASB: a study on the impact on M&A activity. Paper presented at the American Accounting Association Annual Meeting, Anaheim, CA.

Ribeiro, H. R., & Crowther, D. E. (2008c). M&A accounting and activity: a perspective from the U.S.A. to the U.K. Paper presented at the British Accounting Association Annual Conference, Blackpool, United Kingdom.

Rice, J. A. (1995). Mathematical Statistics and Data Analysis (2nd ed.). Belmont, CA: Duxbury Press.

404

Riper, R. V. (1994). Setting Standards for Financial Reporting: FASB and the Struggle for Control of a Critical Process. Westport, CT: Quorum Books

Ripley, W. Z. (1927). Main Street and Wall Street. Boston, MA: Little, Brown & Co. Reprinted in 1972 by Scholars Book Company, Houston, TX.

Roberts, C., Weetman, P., & Gordon, P. (1998). International Financial Accounting: A Comparative Approach (2nd ed.). Essex, UK: Financial Times / Prentice Hall. Pearson Education Limited.

Robinson, J. R., & Shane, P. B. (1990). Acquisition accounting method and bid premium for target firms. The Accounting Review, 65(1), 25-48.

Roll, R. (1986). The hubris hypothesis of corporate takeovers. Journal of Business, 59(2), 197-216.

Rosen, R. J. (2006). Merger momentum and investor sentiment: The stock market reaction to merger announcements. Journal of Business, 79, 987- -1017.

Rosenfeld, P., & Rubin, S. (1985). Contemporary issues in consolidation and the equity method. Journal of Accountancy, 159(6), 94-97.

Ross, S. A., Westerfield, R. W., & Jaffe, J. F. (1996). Corporate Finance (4th ed.). New York: McGraw-Hill, Inc.

Ryan, B., Scapens, R. W., & Theobald, M. (2002). Research Methods and Methodology in Finance and Accounting (2nd ed.). London, UK: Cengage Learning Business Press.

Saltelli, A., Tarantola, S., Campolongo, F., & Ratto, M. (2004). Sensitivity Analysis in Practice: A Guide to Assessing Scientific Models. Chichester, UK: John Wiley & Sons, Ltd.

Sapienza, S. R. (1961). Distinguishing between purchase and pooling. Journal of Accountancy, 111(6), 35-40.

405

Sapienza, S. R. (1962). Pooling theory and practice in business combinations. The Accounting Review, 37(2), 263-278.

Sapienza, S. R. (1964a). Book review: A critical study of accounting for business combinations - Wyatt, A. R. The Accounting Review, 39(1), 230-231.

Sapienza, S. R. (1964b). An examination of AICPA Research Study No. 5 - Standards for pooling. The Accounting Review, 39(3), 582-590.

SAS Institute Inc. (2003). SAS for Windows (Version 9.1) [TS Level 1M3]. Cary, NC: SAS Institute Inc.

SAS Institute Inc. (2004). SAS OnlineDoc® 9.1.3. Cary, NC: SAS Institute Inc.

Savin, N. E., & White, K. J. (1977). The Durbin-Watson test for serial correlation with extreme sample sizes or many regressors. Econometrica, 45(8), 1989-1996.

Schattke, R. W. (1972). An analysis of Accounting Principles Board Statement No. 4. The Accounting Review, 47(2), 233-244.

Scherer, F. M., & Ross, D. (1990). Industrial Market Structure and Economic Performance (3rd ed.). Boston, MA: Houghton Mifflin.

Schroeder, R. H. (1999). Deathwatch for pooling accounting. Bank Accounting & Finance, 12(4), 7-12.

Schultz, S. M., & Hollister, J. (2003). Lobbying FASB on accounting for investments. The Journal of Applied Business Research, 19(2), 85-103.

Seber, G. A. F., & Wild, C. J. (1989). Nonlinear Regression. New York: John Wiley & Sons, Inc.

Securities and Exchange Commission. (1973). Statement of Policy on the Establishment and Improvement of Accounting Principles and Standards, Accounting Series Release No. 150. Washington, DC: U.S. Government Printing Office.

406

Securities and Exchange Commission. (1978). Adoption of Requirements for Financial Accounting and Reporting Practices for Oil and Gas Producing Activities, Accounting Series Release No. 253. Washington, DC: U.S. Government Printing Office.

Securities and Exchange Commission. (2000). Mini-Tender Offers. Retrieved 17 May 2006, http://www.sec.gov/answers/tender.htm

Securities and Exchange Commission. (2003). Tender Offer. Retrieved 17 May 2006, http://www.sec.gov/answers/tender.htm

Segall, J. (1968). Merging for fun and profit. Industrial Management Review, 9(4), 17-30.

Seidler, L. J. (1984). No accounting for the FASB's draft on recognition and measurement. Accounting Issues, Bear Sterns & Co., 13-13.

Sevin, S., & Schroeder, R. (2005). Earnings management: evidence from SFAS No. 142 reporting. Managerial Auditing Journal, 20(1), 47-54.

Sharpe, W. F. (1964). Capital asset prices: a theory of market equilibrium under conditions of risk. Journal of Finance, 19(September), 425-442.

Sharpe, W. F. (1970). Portfolio Theory and Capital Markets. New York: McGraw-Hill, Inc.

Shleifer, A., & Vishny, R. W. (1990). The takeover wave of the 1980s. Science, 249(4971), 745–749.

Shleifer, A., & Vishny, R. W. (1991). Takeovers in the '60s and the '80s: evidence and implications. Strategic Management Journal, 12(Special Issue S2), 51-59.

Shleifer, A., & Vishny, R. W. (2003). Stock market driven acquisitions. Journal of Financial Economics, 70(3), 295-311.

Shughart II, W. F., & Tollison, R. D. (1984). The random character of merger activity. The Rand Journal of Economics, 15(4), 500-509.

407

Singer, E., Hippler, H.-J., & Schwarz, N. (1992). Confidentiality assurances in surveys: reassurance or threat? International Journal of Public Opinion Research, 4(3), 256-268.

Singh, A. (1971). Takeovers: Their Relevance to the Stock Market and the Theory of the Firm. Cambridge: Cambridge University Press.

Sirower, M. L. (1997). The Synergy Trap: How Companies Lose the Acquisition Game. New York: The Free Press.

Smith, A. J. (1981). The SEC "reversal" of FASB Statement No. 19: an investigation of information effects. Journal of Accounting Research, 19(Supplement: Studies on Standardization of Accounting Practices: An Assessment of Alternative Institutional Arrangements), 174-211.

Smith, C. W., & Watts, R. (1982). Incentive and tax effects of U.S. executive compensation plans. Australian Journal of Management, 7(2), 139-157.

Smith, E. E. (1999). Pooling ruling could dampen economy, August 24. Boston, MA: Boston Globe.

Snavely, H. J. (1968). "Pooling" is good accounting. Financial Analysts Journal, 24(6), 85-89.

Solomons, D. (1978). The politicization of accounting. Journal of Accountancy, 146(5), 65-72.

Solomons, D. (1991). Accounting and social change: a neutralist view. Accounting, Organizations and Society, 16(3), 287-295.

Spacek, L. (1958). The need for an accounting court. The Accounting Review, 33(3), 368-379.

Sprague, C. E. (1908). The Philosophy of Accounts: Privately published. Reissued in 1972 by The Scholars Book Company, Lawrence, KS.

408

Sprouse, R. T. (1971). Balance sheet: embodiment of the most fundamental elements of accounting theory. In Foundations of Accounting Theory (pp. 90-104). Gainesville, FL: University of Florida Press.

StatPoint Inc. (2005). The User’s Guide to STATGRAPHICS® Centurion XV. United States of America: StatPoint, Inc.

StatPoint Inc. (2006). STATGRAPHICS® Centurion XV (Version 15.1.02) [Professional]. United States of America: StatPoint, Inc.

StatSoft Inc. (2001). STATISTICA® (data analysis software system) (Version 6). United States of America: StatSoft, Inc.

Staubus, G. J. (1972). An analysis of APB Statement No. 4. Journal of Accountancy, 133(2), 36-43.

Staubus, G. J. (1985). An induced theory of accounting measurement. The Accounting Review, 60(1), 53-75.

Staubus, G. J. (2003). An accountant’s education. The Accounting Historians Journal, 30(1), 155-196.

Stearns, L. B., & Allan, K. D. (1996). Economic behavior in institutional environments: the corporate merger wave of the 1980s. American Sociological Review, 61(4), 699-718.

Steiner, P. O. (1975). Mergers: Motives, Effects, Policies. Ann Arbor, MI: University of Michigan Press.

Stevens, J. P. (2002). Applied Multivariate Statistics for the Social Sciences (4th ed.). Mahwah, NJ: Lawrence Erlbaum Associates.

Stevens, J. P. (2007). Intermediate Statistics: A Modern Approach (3th ed.). New York, NY: Lawrence Erlbaum Associates.

Stigler, G. J. (1950). Monopoly and oligopoly by merger. The American Economic Review, 40(2), 23-34.

409

Sudarsanam, P. S. (1995). The Essence of Mergers and Acquisitions: Prentice Hall Europe.

Sudarsanam, P. S. (2003). Creating Value from Mergers and Acquisitions. Harlow, Essex: Pearson Education.

Sutton, T. G. (1984). Lobbying of accounting standard setting bodies in the UK and the USA: a downsian analysis. Accounting, Organizations and Society, 9(1), 81-95.

Tarca, A. (2004). International convergence of accounting practices: choosing between IAS and US GAAP. Journal of International Financial Management and Accounting, 15(1), 60-91.

Testimony of Alain J. Hanover, CEO, InCert Software Corporation, to the Senate Committee on Banking, Housing and Urban Development on behalf of The Massachusetts Software and Internet Council on Pooling of Interest Methods of Accounting, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

Testimony of Kimberly J. Pinter, Director, Corporate Finance and Tax National Association of Manufacturers On behalf of the National Association of Manufacturers before the Committee on Banking, Housing, and Urban Affairs United States Senate On Business Combinations and Intangible Assets, Senate Banking Committee, Hearing on Pooling Accounting Sess. (2000).

Theil, H. (1966). Applied Economic Forecasting. Amsterdam, Netherlands: North Holland.

Thomson Financial - Sikora, M. (2002). 2001 M&A Profile. Mergers & Acquisitions, 37(2), 13-41.

Thomson Financial. (2006). SDC Platinum database: Thomson Financial.

Thomson Financial. (2009). SDC Platinum database: Thomson Reuters.

410

Thorp, W. L. (1941). The merger movement. In T. N. E. Committee (Ed.), The Structure of Industry, Monograph No. 27, Part 3 (pp. 231-234). Washington, DC: U.S. Government Printing Office.

Tinker, T. (1988). Panglossian accounting theories: the science of apologising in style. Accounting, Organizations & Society, 13(2), 165-189.

Tinker, T., Lehman, C., & Neimark, M. (1988). Marginalizing the public interest: A critical look at recent social accounting history In K. R. Ferris (Ed.), Behavioral Accounting Research: A Critical Analysis (pp. 117- -143). Columbus, OH: Century VII Publishing Company.

Tinker, T., Merino, B. D., & Neimark, M. D. (1982). The normative origins of positive theories: ideology and accounting thought. Accounting, Organizations & Society, 7(2), 167-200.

Tinker, T., & Puxty, A. G. (1995). Policing Accounting Knowledge : The Market for Excuses Affair. Princeton, N.J.: M. Wiener Publishers.

Town, R. J. (1992). Merger waves and the structure of merger and acquisition time-series. Journal of Applied Econometrics, 7(Supplement: Special Issue on Nonlinear Dynamics and Econometrics), S83-S100.

Toxvaerd, F. (2008). Strategic merger waves: a theory of musical chairs. Journal of Economic Theory, 140(1), 1-26.

U.S. Census Bureau. (2008). Statistics about Business Size: Statistics of U.S. Businesses. van Riper, R. (1994). Setting Standards for Financial Reporting: FASB and the Struggle for Control of a Critical Process. Westport, CT: Quorum Books. van Wegberg, M. (1994). Why do mergers occur in waves? Business cycles, bandwagons and the merger for market power paradox. METEOR Discussion Paper.

411

Vatter, W. J. (1963). Postulates and principles. Journal of Accounting Research, 2(3), 179-197.

Vincent, L. (1997). Equity valuation implications of purchase versus pooling accounting. Journal of Financial Statement Analysis, 2(4), 5-19.

Visser, P. S., Krosnick, J. A., Marquette, J., & Curtin, M. (1996). Mail surveys for election forecasting? An evaluation of the Colombia dispatch poll. Public Opinion Quarterly, 60(2), 181-227.

Wakefield, B. R. (1970). APB is on wrong track: banker's view. Financial Analysts Journal, 26(4), 33-36.

Wal-Mart Stores Inc. (2002). SEC Filing: Form 10-K Pursuant to Section 13 or 15(d) of the Securities Exchange Act of 1934 - For the year ended January 31, 2002. Retrieved 2005 (several accessing days), http://www.edgar-online.com

Walter, J. R. (1999). Pooling or purchase: a merger mystery. Federal Reserve Bank of Richmond Economic Quarterly, 85(1), 27-46.

Watts, R. L. (1977). Corporate financial statements, a product of the market and political processes. Australian Journal of Management, 2(1), 53-75.

Watts, R. L. (2006). What has the invisible hand achieved? Accounting and Business Research, 36(Special Issue: International Accounting Policy Forum), 51-61.

Watts, R. L., & Zimmerman, J. L. (1978). Towards a positive theory of the determination of accounting standards. The Accounting Review, 53(1), 112-134.

Watts, R. L., & Zimmerman, J. L. (1979). The demand for and supply of accounting theories: the market for excuses. The Accounting Review, 54(2), 273-305.

412

Watts, R. L., & Zimmerman, J. L. (1986). Positive Accounting Theory. Englewood Cliffs, NJ: Prentice-Hall.

Watts, R. L., & Zimmerman, J. L. (1990). Positive accounting theory: a ten year perspective. The Accounting Review, 65(1), 131-156.

Weber, J. P. (2004). Shareholder wealth effects of pooling-of-interests accounting: evidence from the SEC's restriction on share repurchases following pooling transactions. Journal of Accounting & Economics, 37(1), 39-57.

Website of Stanford University (Online). (2007). Critical values for the Durbin- -Watson test. Retrieved 30 June 2007, http://www.stanford.edu/~clint/bench/dwcrit.htm

Weisberg, S. (1985). Applied Linear Regression (2nd ed.). New York: John Wiley & Sons, Inc.

Wells, M. C. (1976). A revolution in accounting thought. The Accounting Review, 51(3), 471-482.

Wells, M. C. (1977). A revolution in accounting thought? A reply. The Accounting Review, 52(3), 748-750.

Weston, J. F. (1953). The Role of Mergers in the Growth of Large Firms. Berkeley & Los Angeles, CA: University of California Press.

Weston, J. F. (1961). The Role of Mergers in the Growth of Large Firms (2nd printing ed.). Berkeley & Los Angeles, CA: University of California Press.

Weston, J. F. (2001). Mergers & Acquisitions. Blacklick, OH: McGraw-Hill Professional Book Group.

Weston, J. F., Chung, K. S., & Hoag, S. E. (1990). Takeovers, Restructuring and Corporate Governance (1st ed.). Englewood Cliffs, NJ: Prentice Hall.

413

Weston, J. F., Chung, K. S., & Siu, J. A. (1998). Takeovers, Restructuring and Corporate Governance (2nd ed.). Upper Saddle River, NJ: Prentice Hall.

Weston, J. F., & Mansinghka, S. K. (1971). Tests of the efficiency performance of conglomerate firms. The Journal of Finance, 26(4), 919-936.

Weston, J. F., Mitchell, M. L., & Mulherin, J. H. (2004). Takeovers, Restructuring and Corporate Governance (4th ed.). Upper Saddle River, NJ: Pearson Education.

White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48(4), 817-838.

Whittington, G. (1980). Pioneers of income measurement and price-level accounting: a review article. Accounting and Business Research, 10(1), 158-173.

Whittington, G. (1987). Positive accounting: a review article. Accounting and Business Research, 17(6), 327-336.

Wong, P., & O’Sullivan, N. (2001). The determinants and consequences of abandoned takeovers. Journal of Economic Surveys, 15(2), 145-186.

Wyatt, A. R. (1963). Summary from ARS No. 5 - A critical study of accounting for business combinations. Journal of Accountancy, 116(1), 65-67.

Wyatt, A. R. (1965). Accounting for business combinations: what next? The Accounting Review, 40(3), 527-535.

Wyatt, A. R. (1967). Discussion of purchase versus pooling of interests: the search for a predictor. Journal of Accounting Research, Supplement to Vol. 5(Empirical Research in Accounting: Selected Studies), 210-212.

Wyatt, A. R. (1972). Inequities in accounting for business combinations. Financial Executive, 40(12), 28-35.

414

Wyatt, A. R. (1977). The economic impact of financial accounting standards. In 1976 Accounting Research Convocation. Emerging Issues (pp. 129-138). AL: University of Alabama.

Wyatt, A. R. (1986). Standard setting: processes and politics. FASB Viewpoints, 1-3.

Wyatt, A. R. (1990). Accounting standards: conceptual or political? Accounting Horizons, 4(3), 83-88.

Yang, C., & Qiu-sheng, Z. (2006). Empirical Study on Choices between M&A and Strategic Alliance. International Conference on Management Science and Engineering proceedings, 931-935.

Yang, J. M. (1927). Goodwill and Other Intangibles: Their Significance and Treatment in Accounts. New York: Ronald Press Company.

Zeff, S. A. (1978). The rise of economic consequences. Journal of Accountancy, 146(6), 56-63.

Zeff, S. A. (1993). The politics of accounting standards. Economia Aziendale, 12(August), 125-142.

Zeff, S. A. (1995). A perspective on the U.S. public/private-sector approach to the regulation of financial reporting. Accounting Horizons, 9(1), 52-70.

Zeff, S. A. (2000). Henry Rand Hatfield: Humanist, Scholar, and Accounting Educator, Studies in the Development of Accounting Thought. Stamford, CT: JAI Press, Inc.

Zeff, S. A. (2001). The work of the special committee on research program. Accounting Historians Journal, 28(2), 141-186.

Zeff, S. A. (2002). "Political" lobbying on proposed standards: a challenge to the IASB. Accounting Horizons, 16(1), 43-54.

Zeff, S. A. (2003a). How the U.S. Accounting Profession Got Where It Is Today: Part I. Accounting Horizons, 17(3), 189-205.

415

Zeff, S. A. (2003b). How the U.S. Accounting Profession Got Where It Is Today: Part II. Accounting Horizons, 17(4), 267-286.

Zeff, S. A. (2004). Evolution of US Generally Accepted Accounting Principles (GAAP). Beijing, People's Republic of China: International Symposium on Accounting Standards.

Zeff, S. A. (2007). The SEC rules historical cost accounting: 1934 to the 1970s. Accounting and Business Research, 37(International Accounting Policy Forum), 1-14.

416

Appendix A S&P 500 Companies List as of 31 December 2004

Company Industry Country Fiscal Year 3M Company Industrials USA Dec-30 Abbott Labs Health Care USA Dec-30 ACE Limited Financials Bermuda Dec-31 ADC Telecommunications Information Technology USA Dec-30 Adobe Systems Information Technology USA Nov-30 Advanced Micro Devices Information Technology USA Dec-30 AES Corp. Utilities USA Dec-31 Aetna Inc. Health Care USA Dec-31 Affiliated Computer Information Technology USA Jun-30 AFLAC Inc. Financials USA Dec-31 Agilent Technologies Information Technology USA Oct-31 Air Products & Chemicals Materials USA Sep-30 Alberto-Culver Consumer Staples USASep-30 Albertson's Consumer Staples USAJan-31 Alcoa Inc. Materials USA Dec-31 Allegheny Energy Utilities USA Dec-31 Allegheny Technologies Inc. Materials USA Dec-31 Allergan Inc. Health Care USA Dec-31 Allied Waste Industries Industrials USA Dec-31 Allstate Corp. Financials USA Dec-31 ALLTEL Corp. Telecommunication Services USA Dec-31 Altera Corp. Information Technology USA Dec-31 Altria Group, Inc. Consumer Staples USA Dec-31 Ambac Financial Group Financials USA Dec-31 Amerada Hess Energy USA Dec-31 Ameren Corporation Utilities USA Dec-31 American Electric Power Utilities USA Dec-31 American Express Financials USA Dec-31 American Int'l. Group Financials USA Dec-31 American Power Conversion Industrials USA Dec-31 American Standard Industrials USA Dec-31 AmerisourceBergen Corp. Health Care USA Sep-30 Amgen Health Care USADec-31 AmSouth Bancorporation Financials USA Dec-31 Anadarko Petroleum Energy USA Dec-31

417

Analog Devices Information Technology USA May-05 Andrew Corp. Information Technology USA Sep-30 Anheuser-Busch Consumer Staples USADec-31 Aon Corp. Financials USA Dec-31 Apache Corp. Energy USA Dec-31 Apartment Investment & Mgmt. Financials USA Dec-31 Apollo Group Industrials USA Aug-31 Apple Computer Information Technology USA Sep-30 Applera Corp-Applied Biosystems Grp. Health Care USA Jun-30 Applied Materials Information Technology USA Oct-26 Applied Micro Circuits Information Technology USA Mar-31 Archer-Daniels-Midland Consumer Staples USA Jun-30 Archstone-Smith Trust Financials USA Dec-31 Ashland Inc. Energy USA Sep-30 AT&T Corp. Telecommunication Services USA Dec-31 Autodesk Inc. Information Technology USA Jan-31 Automatic Data Processing Inc. Information Technology USA Jun-30 AutoNation, Inc. Consumer Discretionary USA Dec-31 AutoZone Inc. Consumer Discretionary USA Aug-28 Avaya Inc. Information Technology USA Sep-30 Avery Dennison Corp. Industrials USA Dec-31 Avon Products Consumer Staples USA Dec-31 Baker Hughes Energy USA Dec-31 Ball Corp. Materials USA Dec-31 Bank of America Corp. Financials USA Dec-31 Bank of New York Financials USA Dec-31 Bard (C.R.) Inc. Health Care USA Dec-31 Bausch & Lomb Health Care USA Dec-26 Baxter International Inc. Health Care USA Dec-31 BB&T Corporation Financials USA Dec-31 Bear Stearns Cos. Financials USA Nov-30 Becton, Dickinson Health Care USA Sep-30 Bed Bath & Beyond Consumer Discretionary USA Feb-28 BellSouth Telecommunication Services USA Dec-31 Bemis Company Materials USA Dec-31 Best Buy Co., Inc. Consumer Discretionary USA Feb-28 Big Lots Inc. Consumer Discretionary USA Jan-31 Biogen Idec Inc. Health Care USA Dec-31 Biomet Inc. Health Care USA May-31 BJ Services Energy USA Sep-30

418

Black & Decker Corp. Consumer Discretionary USA Dec-31 Block H&R Industrials USA Apr-30 BMC Software Information Technology USA Mar-31 Boeing Company Industrials USA Dec-31 Boston Scientific Health Care USA Dec-31 Bristol-Myers Squibb Health Care USA Dec-31 Broadcom Corporation Information Technology USA Dec-31 Brown-Forman Corp. Consumer Staples USA Apr-30 Brunswick Corp. Consumer Discretionary USA Dec-31 Burlington Northern Santa Fe C. Industrials USA Dec-31 Burlington Resources Energy USA Dec-31 Calpine Corp. Utilities USA Dec-31 Campbell Soup Consumer Staples USA Jul-29 Capital One Financial Financials USA Dec-31 Cardinal Health Inc. Health Care USA Jun-30 Caremark Rx Health Care USA Dec-31 Carnival Corp. Consumer Discretionary USA Nov-30 Caterpillar Inc. Industrials USA Dec-31 Cendant Corporation Industrials USA Dec-31 CenterPoint Energy Utilities USA Dec-31 Centex Corp. Consumer Discretionary USA Mar-31 Century Telephone Telecommunication Services USA Dec-31 Charles Schwab Financials USA Dec-31 ChevronTexaco Corp. Energy USA Dec-31 Chiron Corp. Health Care USA Dec-31 Chubb Corp. Financials USA Dec-31 Ciena Corp. Information Technology USA Oct-31 CIGNA Corp. Health Care USA Dec-31 Cincinnati Financial Financials USA Dec-31 CINergy Corp. Utilities USA Dec-31 Cintas Corporation Industrials USA May-31 Circuit City Group Consumer Discretionary USA Feb-28 Cisco Systems Information Technology USA Jul-31 CIT Group Financials USA Dec-31 Citigroup Inc. Financials USA Dec-31 Citizens Communications Telecommunication Services USA Dec-31 Citrix Systems Information Technology USA Dec-31 Clear Channel Communications Consumer Discretionary USA Dec-31 Clorox Co. Consumer Staples USA Jun-30 CMS Energy Utilities USA Dec-31

419

Coach Inc. Consumer Discretionary USA Jun-30 Coca Cola Co. Consumer Staples USA Dec-31 Coca-Cola Enterprises Consumer Staples USA Dec-31 Colgate-Palmolive Consumer Staples USA Dec-31 Comcast Corp. Consumer Discretionary USA Dec-31 Comerica Inc. Financials USA Dec-31 Compass Bancshares Financials USA Dec-31 Computer Associates Int’l. Information Technology USA Mar-31 Computer Sciences Corp. Information Technology USA Apr-02 Compuware Corp. Information Technology USA Mar-31 Comverse Technology Information Technology USA Jan-31 ConAgra Foods Inc. Consumer Staples USA May-31 ConocoPhillips Energy USA Dec-31 Consolidated Edison Utilities USA Dec-31 Constellation Energy Group Utilities USA Dec-31 Convergys Corp. Information Technology USA Dec-31 Cooper Industries Ltd. Industrials USA Dec-31 Cooper Tire & Rubber Consumer Discretionary USA Dec-31 Coors (Adolph) Consumer Staples USA Dec-28 Corning Inc. Information Technology USA Dec-28 Costco Co. Consumer Staples USA Aug-30 Countrywide Financial Corp. Financials USA Dec-31 CSX Corp. Industrials USA Dec-31 Cummins Inc. Industrials USA Dec-31 CVS Corp. Consumer Staples USA Dec-31 Dana Corp. Consumer Discretionary USA Dec-31 Danaher Corp. Industrials USA Dec-31 Darden Restaurants Consumer Discretionary USA May-26 Deere & Co. Industrials USA Oct-31 Dell Inc. Information Technology USA Jan-29 Delphi Corporation Consumer Discretionary USA Dec-31 Delta Air Lines Industrials USA Dec-31 Devon Energy Corp. Energy USA Dec-31 Dillard Inc. Consumer Discretionary USA Jan-05 Dollar General Consumer Discretionary USA Jan-31 Dominion Resources Utilities USA Dec-31 Donnelley (R.R.) & Sons Industrials USA Dec-31 Dover Corp. Industrials USA Dec-31 Dow Chemical Materials USA Dec-31 Dow Jones & Co. Consumer Discretionary USA Dec-31

420

DTE Energy Co. Utilities USA Dec-31 Du Pont (E.I.) Materials USA Dec-31 Duke Energy Utilities USA Dec-31 Dynegy Inc. Utilities USA Dec-31 E*Trade Financial Corp. Financials USA Dec-31 Eastman Chemical Materials USA Dec-31 Eastman Kodak Consumer Discretionary USA Dec-31 Eaton Corp. Industrials USA Dec-31 eBay Inc. Consumer Discretionary USA Dec-31 Ecolab Inc. Materials USA Dec-31 Edison Int'l. Utilities USA Dec-31 El Paso Corp. Energy USA Dec-31 Electronic Arts Information Technology USA Mar-31 Electronic Data Systems Information Technology USA Dec-31 EMC Corp. Information Technology USA Dec-31 Emerson Electric Industrials USA Sep-30 Engelhard Corp. Materials USA Dec-31 Entergy Corp. Utilities USA Dec-31 EOG Resources Energy USA Dec-31 Equifax Inc. Industrials USA Dec-31 Equity Office Properties Financials USA Dec-31 Equity Residential Financials USA Dec-31 Exelon Corp. Utilities USA Dec-31 Express Scripts Health Care USA Dec-31 Exxon Mobil Corp. Energy USA Dec-31 Family Dollar Stores Consumer Discretionary USA Aug-29 Fannie Mae Financials USA Dec-31 Federal Home Loan Mtg. Financials USA Dec-31 Federated Dept. Stores Consumer Discretionary USA Jan-31 Federated Investors Inc. Financials USA Dec-31 FedEx Corporation Industrials USA May-31 Fifth Third Bancorp Financials USA Dec-31 First Data Information Technology USA Dec-31 First Horizon National Financials USA Dec-31 FirstEnergy Corp. Utilities USA Dec-31 FIserv Inc. Information Technology USA Dec-31 Fisher Scientific Health Care USA Dec-31 Fluor Corp. Industrials USA Dec-31 Ford Motor Consumer Discretionary USA Dec-31 Forest Laboratories Health Care USA Mar-31

421

Fortune Brands Inc. Consumer Discretionary USA Dec-31 FPL Group Utilities USA Dec-31 Franklin Resources Financials USA Sep-30 Freeport-McMoran Cp & Gld Materials USA Dec-31 Freescale Semiconductor Inc. Information Technology USA Dec-31 Gannett Co. Consumer Discretionary USA Dec-31 Gap (The) Consumer Discretionary USA Jan-31 Gateway Inc. Information Technology USA Dec-31 General Dynamics Industrials USA Dec-31 General Electric Industrials USA Dec-31 General Mills Consumer Staples USA May-25 General Motors Consumer Discretionary USA Dec-31 Genuine Parts Consumer Discretionary USA Dec-31 Genzyme Corp. Health Care USA Dec-31 Georgia-Pacific Group Materials USA Dec-28 Gilead Sciences Health Care USA Dec-31 Gillette Co. Consumer Staples USA Dec-31 Golden West Financial Financials USA Dec-31 Goldman Sachs Group Financials USA Nov-28 Goodrich Corporation Industrials USA Dec-31 Goodyear Tire & Rubber Consumer Discretionary USA Dec-31 Grainger (W.W.) Inc. Industrials USA Dec-31 Great Lakes Chemical Materials USA Dec-31 Guidant Corp. Health Care USA Dec-17 Halliburton Co. Energy USA Dec-31 Harley-Davidson Consumer Discretionary USA Dec-31 Harrah's Entertainment Consumer Discretionary USA Dec-31 Hartford Financial Svc. Gp. Financials USA Dec-31 Hasbro Inc. Consumer Discretionary USA Dec-27 HCA Inc. Health Care USA Dec-31 Health Management Assoc. Health Care USA Sep-30 Heinz (H.J.) Consumer Staples USA Apr-30 Hercules Inc. Materials USA Dec-31 Hershey Foods Consumer Staples USA Dec-31 Hewlett-Packard Information Technology USAOct-31 Hilton Hotels Consumer Discretionary USA Dec-31 Home Depot Consumer Discretionary USA Jan-31 Honeywell Int'l Inc. Industrials USA Dec-31 Hospira Inc. Health Care USA Dec-31 Humana Inc. Health Care USA Dec-31

422

Huntington Bancshares Financials USA Dec-31 Illinois Tool Works Industrials USA Dec-31 IMS Health Inc. Health Care USA Dec-31 Ingersoll-Rand Co. Ltd. Industrials USA Dec-31 Intel Corp. Information Technology USA Dec-31 International Bus. Machines Information Technology USA Dec-31 International Flav/Frag Materials USA Dec-31 International Game Technology Consumer Discretionary USA Sep-30 International Paper Materials USA Dec-31 Interpublic Group Consumer Discretionary USA Dec-31 Intuit Inc. Information Technology USA Jul-31 ITT Industries Inc. Industrials USA Dec-31 Jabil Circuit Information Technology USA Aug-31 Janus Capital Group Financials USA Dec-31 JDS Uniphase Corp Information Technology USA Jun-30 Jefferson-Pilot Financials USA Dec-31 Johnson & Johnson Health Care USA Mar-05 Johnson Controls Consumer Discretionary USA Sep-30 Jones Apparel Group Consumer Discretionary USA Dec-31 JPMorgan Chase & Co. Financials USA Dec-31 KB Home Consumer Discretionary USA Nov-30 Kellogg Co. Consumer Staples USA Dec-31 Kerr-McGee Energy USA Dec-31 KeyCorp Financials USADec-31 Keyspan Energy Utilities USA Dec-31 Kimberly-Clark Consumer Staples USA Dec-31 Kinder Morgan Energy USA Dec-31 King Pharmaceuticals Health Care USA Dec-31 KLA-Tencor Corp. Information Technology USA Jun-30 Knight-Ridder Inc. Consumer Discretionary USA Dec-27 Kohl's Corp. Consumer Discretionary USA Jan-30 Kroger Co. Consumer Staples USA Feb-05 L-3 Communications Holdings Industrials USA Dec-31 Laboratory Corp. of America Holding Health Care USA Dec-31 Leggett & Platt Consumer Discretionary USA Dec-31 Lehman Bros. Financials USA Nov-30 Lexmark Int'l Inc. Information Technology USA Dec-31 Lilly (Eli) & Co. Health Care USA Dec-31 Limited Brands Inc. Consumer Discretionary USA Jan-31 Lincoln National Financials USA Dec-31

423

Linear Technology Corp. Information Technology USA Jun-27 Liz Claiborne, Inc. Consumer Discretionary USA Dec-28 Lockheed Martin Corp. Industrials USA Dec-31 Loews Corp. Financials USA Dec-31 Louisiana Pacific Materials USA Dec-31 Lowe's Cos. Consumer Discretionary USA Jan-29 LSI Logic Information Technology USA Dec-31 Lucent Technologies Information Technology USA Sep-30 M&T Bank Corp. Financials USA Dec-31 Manor Care Inc. Health Care USA Dec-31 Marathon Oil Corp. Energy USA Dec-31 Marriott Int'l. Consumer Discretionary USA Dec-31 Marsh & McLennan Financials USA Dec-31 Marshall & Ilsley Corp. Financials USA Dec-31 Masco Corp. Industrials USA Dec-31 Mattel Inc. Consumer Discretionary USA Dec-31 Maxim Integrated Prod Information Technology USA Dec-15 May Dept. Stores Consumer Discretionary USA Jan-05 Maytag Corp. Consumer Discretionary USA Dec-31 MBIA Inc. Financials USA Dec-31 MBNA Corp. Financials USA Dec-31 McCormick & Co. Consumer Staples USA Nov-30 McDonald's Corp. Consumer Discretionary USA Dec-31 McGraw-Hill Consumer Discretionary USADec-31 McKesson Corp. Health Care USA Mar-31 MeadWestvaco Corporation Materials USA Dec-31 Medco Health Solutions Inc. Health Care USA Dec-29 MedImmune Inc. Health Care USA Dec-31 Medtronic Inc. Health Care USA Apr-30 Mellon Bank Corp. Financials USA Dec-31 Merck & Co. Health Care USA Dec-31 Mercury Interactive Information Technology USA Dec-31 Meredith Corp. Consumer Discretionary USA Jun-30 Merrill Lynch Financials USA Dec-31 MetLife Inc. Financials USA Dec-31 MGIC Investment Financials USA Dec-31 Micron Technology Information Technology USA Aug-31 Microsoft Corp. Information Technology USA Jun-30 Millipore Corp. Health Care USA Dec-31 Molex Inc. Information Technology USA Jun-30

424

Monsanto Co. Materials USA Aug-31 Monster Worldwide Industrials USA Dec-31 Moody's Corp. Financials USA Dec-31 Morgan Stanley Financials USA Nov-30 Motorola Inc. Information Technology USA Dec-31 Mylan Laboratories Health Care USA Mar-31 Nabors Industries Ltd. Energy Barbados Dec-31 National City Corp. Financials USA Dec-31 National Semiconductor Information Technology USA May-31 Navistar International Corp. Industrials USA Oct-31 NCR Corp. Information Technology USA Dec-31 Network Appliance Information Technology USA Apr-30 New York Times Consumer Discretionary USA Dec-31 Newell Rubbermaid Co. Consumer Discretionary USA Dec-31 Newmont Mining Corp. (Hldg. Co.) Materials USA Dec-31 News Corporation Consumer Discretionary USA Jun-30 Nextel Communications Telecommunication Services USA Dec-31 NICOR Inc. Utilities USA Dec-31 NIKE Inc. Consumer Discretionary USA May-31 NiSource Inc. Utilities USA Dec-31 Noble Corporation Energy USA Dec-31 Nordstrom Consumer Discretionary USAJan-29 Norfolk Southern Corp. Industrials USA Dec-31 North Fork Bancorporation Financials USA Dec-31 Northern Trust Corp. Financials USA Dec-31 Northrop Grumman Corp. Industrials USA Dec-31 Novell Inc. Information Technology USA Oct-31 Novellus Systems Information Technology USA Dec-31 Nucor Corp. Materials USA Dec-31 NVIDIA Corp. Information Technology USA Jan-25 Occidental Petroleum Energy USA Dec-31 Office Depot Consumer Discretionary USA Dec-26 OfficeMax Inc. Consumer Discretionary USA Dec-31 Omnicom Group Consumer Discretionary USA Dec-31 Oracle Corp. Information Technology USA May-31 PACCAR Inc. Industrials USA Dec-31 Pactiv Corp. Materials USA Dec-31 Pall Corp. Industrials USA Feb-05 Parametric Technology Information Technology USA Sep-30 Parker-Hannifin Industrials USA Jun-30

425

Paychex Inc. Information Technology USA May-31 Penney (J.C.) Consumer Discretionary USA Jan-26 Peoples Energy Utilities USA Sep-30 Pepsi Bottling Group Consumer Staples USA Dec-31 PepsiCo Inc. Consumer Staples USA Dec-28 PerkinElmer Health Care USA Dec-28 Pfizer Inc. Health Care USA Dec-31 PG&E Corp. Utilities USA Dec-31 Phelps Dodge Materials USA Dec-31 Pinnacle West Capital Utilities USA Dec-31 Pitney-Bowes Industrials USA Dec-31 Plum Creek Timber Co. Financials USA Dec-31 PMC-Sierra Inc. Information Technology USA May-05 PNC Bank Corp. Financials USA Dec-31 Power-One Inc. Industrials USA Dec-31 PPG Industries Materials USA Dec-31 PPL Corp. Utilities USA Dec-31 Praxair Inc. Materials USA Dec-31 Principal Financial Group Financials USA Dec-31 Procter & Gamble Consumer Staples USA Jun-30 Progress Energy Inc. Utilities USA Dec-31 Progressive Corp. Financials USA Dec-31 ProLogis Financials USADec-31 Providian Financial Corp. Financials USA Dec-31 Prudential Financial Financials USA Dec-31 Public Serv. Enterprise Inc. Utilities USA Dec-31 Pulte Homes, Inc. Consumer Discretionary USA Dec-31 QLogic Corp. Information Technology USA Mar-28 QUALCOMM Inc. Information Technology USA Sep-30 Quest Diagnostics Health Care USA Dec-31 Qwest Communications Int Telecommunication Services USA Dec-31 RadioShack Corp Consumer Discretionary USA Dec-31 Raytheon Co. Industrials USA Dec-31 Reebok International Consumer Discretionary USA Dec-31 Regions Financial Corp. Financials USA Dec-31 Reynolds American Inc. Consumer Staples USA Dec-31 Robert Half International Industrials USA Dec-31 Rockwell Automation Inc. Industrials USA Sep-30 Rockwell Collins Industrials USA Sep-30 Rohm & Haas Materials USA Dec-31

426

Rowan Cos. Energy USA Dec-31 Ryder System Industrials USA Dec-31 Sabre Holding Corp. Information Technology USA Dec-31 SAFECO Corp. Financials USA Dec-31 Safeway Inc. Consumer Staples USA Dec-28 Sanmina-SCI Corp. Information Technology USA Sep-30 Sara Lee Corp. Consumer Staples USA Jun-30 SBC Communications Inc. Telecommunication Services USA Dec-31 Schering-Plough Health Care USA Dec-31 Schlumberger Ltd. Energy USA Dec-31 Scientific-Atlanta Information Technology USAJun-30 Sealed Air Corp. Materials USA Dec-31 Sears, Roebuck & Co. Consumer Discretionary USA Dec-28 Sempra Energy Utilities USA Dec-31 Sherwin-Williams Consumer Discretionary USADec-31 Siebel Systems Inc. Information Technology USA Dec-31 Sigma-Aldrich Materials USA Dec-31 Simon Property Group Inc. Financials USA Dec-31 SLM Corporation Financials USA Dec-31 Snap-On Inc. Consumer Discretionary USA Dec-31 Solectron Information Technology USAAug-31 Southern Co. Utilities USA Dec-31 Southwest Airlines Industrials USA Dec-31 Sovereign Bancorp Financials USA Dec-31 Sprint Corp. Telecommunication Services USA Dec-31 St Jude Medical Health Care USA Dec-31 St. Paul Travelers Cos. Financials USA Dec-31 Stanley Works Consumer Discretionary USA Dec-31 Staples Inc. Consumer Discretionary USA Jan-27 Starbucks Corp. Consumer Discretionary USA Sep-28 Starwood Hotels & Resorts Consumer Discretionary USA Dec-31 State Street Corp. Financials USA Dec-31 Stryker Corp. Health Care USA Dec-31 Sun Microsystems Information Technology USA Jun-30 SunGard Data Systems Information Technology USA Dec-31 Sunoco Inc. Energy USA Dec-31 SunTrust Banks Financials USA Dec-31 Supervalu Inc. Consumer Staples USA Feb-22 Symantec Corp. Information Technology USA Feb-05 Symbol Technologies Information Technology USA Dec-31

427

Synovus Financial Financials USA Dec-31 Sysco Corp. Consumer Staples USA Jun-30 T. Rowe Price Group Financials USA Dec-31 Target Corp. Consumer Discretionary USA Jan-31 TECO Energy Utilities USA Dec-31 Tektronix Inc. Information Technology USA May-31 Tellabs Inc. Information Technology USA Dec-31 Temple-Inland Materials USA Dec-30 Tenet Healthcare Corp. Health Care USA Dec-31 Teradyne Inc. Information Technology USA Dec-31 Texas Instruments Information Technology USA Dec-31 Textron Inc. Industrials USA Dec-31 Thermo Electron Health Care USA Dec-31 Tiffany & Co. Consumer Discretionary USA Jan-31 Time Warner Inc. Consumer Discretionary USA Dec-31 TJX Companies Inc. Consumer Discretionary USA Jan-27 Torchmark Corp. Financials USA Dec-31 Toys R Us Inc. Consumer Discretionary USA Jan-30 Transocean Inc. Energy USA Dec-31 Tribune Co. Consumer Discretionary USA Dec-27 TXU Corp. Utilities USA Dec-31 Tyco International Industrials Bermuda Sep-30 U.S. Bancorp Financials USA Dec-31 Union Pacific Industrials USA Dec-31 Unisys Corp. Information Technology USA Dec-31 United Health Group Inc. Health Care USA Dec-31 United Parcel Service Industrials USA Dec-31 United States Steel Corp. Materials USA Dec-31 United Technologies Industrials USA Dec-31 Univision Communications Consumer Discretionary USA Dec-31 Unocal Corp. Energy USA Dec-31 UnumProvident Corp. Financials USA Dec-31 UST Inc. Consumer Staples USA Dec-31 V.F. Corp. Consumer Discretionary USA Dec-29 Valero Energy Energy USA Dec-31 Veritas Software Information Technology USA Dec-31 Verizon Communications Telecommunication Services USA Dec-31 Viacom Inc. Consumer Discretionary USA Dec-31 Visteon Corp. Consumer Discretionary USA Dec-31 Vulcan Materials Materials USA Dec-31

428

Wachovia Corp. Financials USA Dec-31 Walgreen Co. Consumer Staples USA Aug-31 Wal-Mart Stores Consumer Staples USA Jan-31 Walt Disney Co. Consumer Discretionary USA Sep-30 Washington Mutual Financials USA Dec-31 Waste Management Inc. Industrials USA Dec-31 Waters Corporation Health Care USA Dec-31 Watson Pharmaceuticals Health Care USA Dec-31 WellPoint Inc. Health Care USA Dec-31 Wells Fargo Financials USA Dec-31 Wendy's International Consumer Discretionary USA Dec-31 Weyerhaeuser Corp. Materials USA Dec-28 Whirlpool Corp. Consumer Discretionary USA Dec-31 Williams Cos. Energy USA Dec-31 Wrigley (Wm) Jr. Consumer Staples USA Dec-31 Wyeth Health Care USADec-31 Xcel Energy Inc. Utilities USA Dec-31 Xerox Corp. Information Technology USA Dec-31 Xilinx Inc. Information Technology USA Mar-29 XL Capital Financials Bermuda Dec-31 XTO Energy Inc. Energy USA Dec-31 Yahoo Inc. Information Technology USA Dec-31 Yum! Brands Inc Consumer Discretionary USA Dec-31 Zimmer Holdings Health Care USA Dec-31 Zions Bancorp Financials USA Dec-31

429

Appendix B Questionnaire Addressed to S&P 500 companies

1. FASB published in June, 2001 SFAS 141 which superseded APB Opinion No. 16. Among many changes about business combinations accounting it banned the pooling of interests method. Did the publication of SFAS 141 contribute or constrain in any way the completion of a M&A? If so, please describe above how, if not just please mark a cross in “no effect”. Contributed Constrained No Effect

2. FASB published in June, 2001 SFAS 142 which superseded APB Opinion No. 17. Among many changes about business combinations accounting it replaced goodwill amortization by impairment tests. Did the publication of SFAS 142 contribute or constrain in any way the completion of a M&A? If so, please describe above how, if not just please mark a cross in “no effect”. Contributed Constrained No Effect

3. How many times has your corporation withdrawn any planned or announced M&A due to the publication of SFAS 141 and/or SFAS 142? (If possible, please indicate the main reasons) SFAS 141 SFAS 142 Both SFAS 141 & 142

4. How many times has your corporation brought forward any M&A activity to try to qualify the deal for pooling of interests method before its ban, avoiding purchase accounting method, effective after July 1, 2001? Please provide further information explaining the main motivations of such action.

5. How would you classify the relevance and impact of SFAS 141 and SFAS 142 new accounting rules to the M&A decision-making and overall activity? What is the importance? (1=Low; 2=Medium; 3=High) 1 2 3 SFAS 141 SFAS 142

430

Appendix C Crosstabulations for Questionnaire Data

Table C.1 Crosstabulation for SFAS 141 and IT & Financials industries

IT & Financials Total 0 1 (IT&F) 0 N 31 17 48 SFAS (no effect) % 59.6 32.6 92.3 141 N 1 3 4 1 % 1.9 5.7 7.6 N 32 20 52 Total % 61.5 38.4 100.0

Test Value Df P-Value Chi-Squared 2.444 1 0.1180 Pearson's R 0.2168 50 0.1226 Contingency Coefficient 0.2119 - - Cramer's V 0.2168 - -

The hypothesis that FAS 141 and IT & Financials industries are independent at the 95% confidence level cannot be rejected.

431

Table C.2 Crosstabulation for SFAS 142 and IT & Financials industries

IT & Financials Total 0 1 (IT&F) 0 N 29 11 40 SFAS (no effect) % 55.7 21.1 76.9 142 N 3 9 12 1 % 5.7 17.3 23.0 N 32 20 52 Total % 61.5 38.4 100.0

Test Value Df P-Value Chi-Squared 8.799 1 0.0030 Pearson's R 0.4114 50 0.0024 Contingency Coefficient 0.3804 - - Cramer's V 0.4114 - -

The hypothesis that FAS 142 and IT & Financials industries are independent at the 95% confidence level cannot be rejected.

432

Table C.3 Crosstabulation for SFAS 141 and IT industry

IT Industry Total 0 1 (IT) 0 N 46 2 48 SFAS (no effect) % 88.4 3.8 92.3 141 N 2 2 4 1 % 3.8 3.8 7.6 N 48 4 52 Total % 92.3 7.6 100.0

Test Value Df P-Value Chi-Squared 10.924 1 0.0009 Pearson's R 0.4583 50 0.0006 Contingency Coefficient 0.4167 - - Cramer's V 0.4583 - -

The hypothesis that FAS 141 and IT industry are independent at the 95% confidence level can be rejected.

433

Table C.4 Crosstabulation for SFAS 142 and IT industry

IT Industry Total 0 1 (IT) 0 N 39 1 40 SFAS (no effect) % 75.0 1.9 76.9 142 N 9 3 12 1 % 17.3 5.7 23.0 N 48 4 52 Total % 92.3 7.6 100.0

Test Value Df P-Value Chi-Squared 6.581 1 0.0103 Pearson's R 0.3558 50 0.0096 Contingency Coefficient 0.3352 - - Cramer's V 0.3558 - -

The hypothesis that FAS 142 and IT industry are independent at the 95% confidence level can be rejected.

434

Table C.5 Crosstabulation for SFAS 141 and Financials industry

Financials Total 0 1 (Fin.) 0 N 33 15 48 SFAS (no effect) % 63.4 28.8 92.3 141 N 3 1 4 1 % 5.7 1.9 7.6 N 36 16 52 Total % 69.2 30.7 100.0

Test Value Df P-Value Chi-Squared 0.068 1 0.7947 Pearson's R -0.0361 50 0.7995 Contingency Coefficient 0.0361 - - Cramer's V 0.0361 - -

The hypothesis that FAS 141 and financial industry are independent at the 95% confidence level cannot be rejected.

435

Table C.6 Crosstabulation for SFAS 142 and Financials industry

Financials Total 0 1 (Fin.) 0 N 30 10 40 SFAS (no effect) % 57.6 19.2 76.9 142 N 6 6 12 1 % 11.5 11.5 23.0 N 36 16 52 Total % 69.2 30.7 100.0

Test Value Df P-Value Chi-Squared 2.708 1 0.0998 Pearson's R 0.2282 50 0.1037 Contingency Coefficient 0.2225 - - Cramer's V 0.2282 - -

The hypothesis that FAS 142 and financial industry are independent at the 95% confidence level cannot be rejected.

436

Appendix D Goodwill and Other Intangible Assets (OIA), and Impact on Diluted EPS, by Industry

Fig. D.1 Weighted average goodwill and OIA, and diluted EPS, for 2000-01

175 170.9

150 142.4 137.6

125 104.8 100.7 100 79.4 75

44.6 53.0 50 42.5 38.3 27.9 30.0 31.5 23.5 25.7 25 15.7 17.4 14.3 9.3 9.3

0 Consumer Consumer Energy Financials Health Care Industrials IT Materials Telecom. Uti l i ti es Discr. Stap. Serv.

Goodwill & OIA added back ($ mil.) EPS (%)

Fig. D.2 Total amounts of goodwill and OIA added back for 2000-01

Energy Utilities Materials 0.8% 2.2% 3.6% Telecom. Serv. 4.2%

IT Health Care 31.1% 5.0%

Consumer Stap. 9.2%

Financials 10.7%

Consumer Discr. Industrials 21.0% 12.3%

437

Appendix E Descriptive Statistics

Table E.1 Descriptive statistics for unadjusted and non-dummy variables Panel A: Monthly Data (36 months) MA WITH SP500 SP500 Av Fed Fed Av IP GDP MKTC Average 548.833 25.1667 1198.02 1205.24 3.95806 3.94537 1011430 9918.82 14184.5 Median 519.5 24.0 1185.78 1187.69 3.885 3.86412 1006950 9896.8 14081.7 Standard deviation 117.881 8.99365 199.342 197.325 2.09499 2.06975 20035.7 111.09 1972.17 Coeff. of variation 21.478% 35.736% 16.6393% 16.3722% 52.9297% 52.4602% 1.98094% 1.12% 13.9037% Standard error 19.6468 1.49894 33.2237 32.8874 0.349165 0.344959 3339.29 18.5151 328.696 Minimum 323.0 11.0 815.28 854.631 1.16 1.24182 979061 9695.6 10254.7 Maximum 810.0 44.0 1517.68 1485.46 6.86 6.79818 1042310 10095.8 17665.4 Range 487.0 33.0 702.4 630.827 5.7 5.55636 63247.0 400.2 7410.75 Lower quartile 467.0 19.0 1063.46 1078.03 1.82 1.75746 996291 9859.5 12987.1 Upper quartile 633.5 31.0 1380.43 1388.49 5.965 6.00192 1031770 10004.5 15966.5 Stnd. skewness 1.53604 0.89542 -0.37673 -0.52742 0.19347 0.13140 0.26936 -0.000039 -0.333135 Stnd. kurtosis -0.42321 -0.60315 -1.25989 -1.325 -2.13074 -2.18567 -1.53519 -0.255029 -0.975409

438

Table E.1 Descriptive statistics for unadjusted and non-dummy variables (continued) Panel B: Weekly Data (156 weeks) MA SP500 SP500 Av Fed Fed Av Average 126.263 1203.95 1205.78 3.91301 3.94314 Median 120.0 1190.38 1192.39 3.87 3.836 Standard deviation 33.8761 196.854 196.87 2.03157 2.03892 Coeff. of variation 26.8298% 16.3507% 16.3272% 51.9182% 51.7081% Standard error 2.71226 15.761 15.7622 0.162655 0.163244 Minimum 46.0 800.58 799.966 1.18 1.212 Maximum 254.0 1527.45 1512.99 6.86 6.702 Range 208.0 726.87 713.028 5.68 5.49 Lower quartile 102.5 1084.8 1084.54 1.74 1.763 Upper quartile 150.5 1376.1 1382.44 5.955 6.007 Stnd. skewness 2.7034 -0.993522 -1.06723 0.241791 0.205342 Stnd. kurtosis 1.74867 -2.53873 -2.61851 -4.38119 -4.41687

439

Table E.1 Descriptive statistics for unadjusted and non-dummy variables (continued) Panel C (1): Daily Data (782 days) Panel C (2): Daily Data pre & post 1st July 2001 MA SP500 Fed MA pre MA post SP500 pre SP500 post Fed pre Fed post Nr of days 782 782 782 391 391 391 391 391 391 Average 24.6 1204.94 3.9361 28.6 20.6 1368.91 1040.97 5.82414 2.04808 Median 23 1188.45 3.845 28 20 1383.62 1083.82 5.99 1.76 Standard deviation 10.214 197.366 2.0388 10.4679 8.17375 100.936 117.933 0.80936 0.72209 Coef. of variation 41.437% 16.379% 51.797% 36.498% 39.637% 7.373% 11.329% 13.896% 35.257% Minimum 0 776.76 1.15 0 0 1103.25 776.76 3.68 1.15 Maximum 78 1527.45 7.03 78 65 1527.45 1234.45 7.03 3.89 Range 78 750.69 5.88 78 65 424.2 457.69 3.35 2.74 Lower quartile 18 1083.82 1.76 21 15 1298.35 916.07 5.48 1.71 Upper quartile 30 1383.62 5.99 35 25 1452.42 1138.65 6.5 2.04 Stnd. skewness 9.0778 -2.2821 0.50428 5.50464 5.90779 -4.50505 -3.30858 -8.5428 11.8291 Stnd. kurtosis 11.174 -5.8323 -9.76242 7.76695 10.8999 -2.25113 -4.84516 0.59247 3.8041

440

Table E.1 Descriptive statistics for unadjusted and non-dummy variables (continued) Panel D: Quarterly Data (60 quarters) MA WITH SP500 SP500 Av Fed Fed Av IP GDP MKTC Average 2047.85 83.6167 1060.89 1058.93 4.19 4.03393 98.1676 9842.07 13128.4 Median 2058.0 82.0 1137.34 1122.41 4.96 4.80412 100.793 9896.8 13764.5 Standard deviation 369.34 35.331 309.57 309.56 1.9376 1.7772 10.6255 1228.06 4406.13 Coeff. of variation 18.03% 42.25% 29.18% 29.23% 46.24% 44.05% 10.82% 12.47% 33.56% Standard error 47.681 4.5613 39.965 39.9641 0.2501 0.22943 1.37174 158.542 568.829 Minimum 1369.0 28.0 444.27 451.169 0.14 0.50532 74.886 7715.1 4969.94 Maximum 2863.0 151.0 1526.75 1497.18 7.07 6.51946 112.285 11727.4 21070.3 Range 1494.0 123.0 1082.48 1046.01 6.93 6.01413 37.3987 4012.3 16100.4 Lower quartile 1782.0 51.5 864.0 873.644 2.25 2.10758 92.0473 8806.05 10449.0 Upper quartile 2286.5 116.5 1284.54 1285.99 5.825 5.43423 105.981 10911.0 16548.2 Stnd. skewness 0.32534 0.50833 -1.6971 -1.81721 -1.30144 -1.88173 -2.30515 -0.42460 -0.89977 Stnd. kurtosis -0.80507 -2.28496 -1.1076 -1.14555 -1.81172 -1.80187 -0.87245 -1.8241 -1.29108

441

Appendix F Pearson/Spearman Correlation Matrixes

Table F.1 Pearson/Spearman correlation matrixes for coefficient estimates Panel A: Monthly observations MA WITH SP500 SP500 Av Fed Fed Av IP GDP MKTC Per Per 2 Per 3 TD ______MA 0.7390 0.8194 0.8021 0.8004 0.7671 0.7968 -0.6926 0.8227 -0.8216 -0.7114 -0.6185 0.3454 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0001 0.0391 WITH 0.7098 0.7521 0.7659 0.7075 0.6765 0.6920 -0.6163 0.7489 -0.7363 -0.6766 -0.6213 0.1513 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0001 0.0000 0.0000 0.0000 0.0001 0.3783 SP500 0.8384 0.7501 0.9885 0.9065 0.9024 0.7115 -0.8552 0.9894 -0.9464 -0.9379 -0.8978 0.0971 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5731 SP500 Av 0.8311 0.7582 0.9876 0.9163 0.9104 0.7150 -0.8634 0.9721 -0.9543 -0.9487 -0.9105 0.0576 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7384 Fed 0.8000 0.6944 0.8879 0.9019 0.9933 0.8562 -0.7906 0.8565 -0.9396 -0.9139 -0.8492 0.1069 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5350 Fed Av 0.7995 0.6550 0.8947 0.9017 0.9539 0.8496 -0.7816 0.8509 -0.9292 -0.9133 -0.8526 0.0744 0.0000 0.0001 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6665 IP 0.7727 0.6702 0.6963 0.7066 0.7681 0.7596 -0.4755 0.6615 -0.7264 -0.6184 -0.5040 0.1189 0.0000 0.0001 0.0000 0.0000 0.0000 0.0000 0.0034 0.0000 0.0000 0.0001 0.0017 0.4899 GDP -0.7049 -0.6276 -0.8645 -0.8846 -0.8568 -0.8311 -0.5600 -0.8455 0.9303 0.9207 0.8968 0.0401 0.0000 0.0002 0.0000 0.0000 0.0000 0.0000 0.0009 0.0000 0.0000 0.0000 0.0000 0.8162 MKTC 0.8301 0.7342 0.9848 0.9735 0.8503 0.8541 0.6438 -0.8373 -0.9146 -0.9062 -0.8705 0.1315 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0001 0.0000 0.0000 0.0000 0.0000 0.4447 Per -0.8273 -0.7209 -0.9403 -0.9418 -0.9229 -0.9009 -0.6862 0.9411 -0.9094 0.9700 0.9195 -0.0256 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8821 Per 2 -0.8273 -0.7209 -0.9403 -0.9418 -0.9229 -0.9009 -0.6862 0.9411 -0.9094 1.0000 0.9862 -0.0107 ___ 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9508 Per 3 -0.8273 -0.7209 -0.9403 -0.9418 -0.9229 -0.9009 -0.6862 0.9411 -0.9094 1.0000 1.0000 0.0035 ___ 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.9837 TD 0.2666 0.0840 0.0535 0.0306 0.1244 0.1137 0.0216 0.0173 0.0513 -0.0133 -0.0133 -0.0133 0.1148 0.6192 0.7514 0.8563 0.4616 0.5010 0.8982 0.9187 0.7614 0.9373 0.9373 0.9373 Upper-right cells and lower-left cells diagonal represent Pearson and Spearman correlation coefficients, respectively. P-Values in italic. Only non-dummy variables are exhibited. 36 monthly observations. 442

Table F.1 Pearson/Spearman correlation matrixes for coefficient estimates (continued)

Panel B: Weekly observations MA SP500 SP500 Av Fed Fed Av Per Per_2 Per_3 TD

MA 0.6395 0.6397 0.6201 0.6161 -0.6681 -0.5792 -0.5065 0.2886 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0003 SP500 0.6682 0.9954 0.8943 0.8950 -0.9425 -0.9360 -0.8975 0.0375 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6420 SP500 Av 0.6710 0.9935 0.9010 0.9003 -0.9458 -0.9398 -0.9011 0.0425 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.5983 Fed 0.6266 0.8864 0.8910 0.9975 -0.9302 -0.9123 -0.8497 0.0591 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.4639 Fed Av 0.6299 0.8815 0.8823 0.9803 -0.9282 -0.9114 -0.8492 0.0403 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6177 Per -0.6797 -0.9388 -0.9399 -0.8998 -0.8947 0.9686 0.9172 -0.0407 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.6143 Per 2 -0.6797 -0.9388 -0.9399 -0.8998 -0.8947 1.0000 0.9861 -0.0279 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.7292 Per 3 -0.6797 -0.9388 -0.9399 -0.8998 -0.8947 1.0000 1.0000 -0.0193 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.8112 TD 0.1950 0.0234 0.0302 0.0834 0.0322 -0.0325 -0.0325 -0.0325 0.0152 0.7711 0.7067 0.2991 0.6882 0.6857 0.6857 0.6857 Upper-right cells and lower-left cells diagonal represent Pearson and Spearman correlation coefficients, respectively. P-Values in italic. Only non-dummy variables are exhibited. 156 weekly observations. 443

Table F.1 Pearson/Spearman correlation matrixes for coefficient estimates (continued)

Panel C: Daily observations MA SP500 Fed Per Per_2 Per_3 Hol E_BoM HS_Ext

MA 0.4104 0.4027 -0.4307 -0.3694 -0.3202 -0.3032 0.3926 -0.2162 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

SP500 0.4304 0.8965 -0.9435 -0.9371 -0.8981 0.0100 0.0032 -0.0506 0.0000 0.0000 0.0000 0.0000 0.0000 0.7791 0.9279 0.1571

Fed 0.4240 0.8819 -0.9271 -0.9096 -0.8469 -0.0036 0.0141 -0.0532 0.0000 0.0000 0.0000 0.0000 0.0000 0.9206 0.6940 0.1369

Per -0.4371 -0.9387 -0.8941 0.9683 0.9166 -0.0021 -0.0004 0.0563 0.0000 0.0000 0.0000 0.0000 0.0000 0.9528 0.9917 0.1153

Per 2 -0.4371 -0.9387 -0.8941 1.0000 0.9860 -0.0037 -0.0004 0.0524 0.0000 0.0000 0.0000 0.0000 0.0000 0.9178 0.9921 0.1431

Per 3 -0.4371 -0.9387 -0.8941 1.0000 1.0000 -0.0045 -0.0004 0.0502 0.0000 0.0000 0.0000 0.0000 0.0000 0.8996 0.9915 0.1602

Hol -0.2693 0.0082 -0.0200 -0.0021 -0.0021 -0.0021 -0.0591 -0.0330 0.0000 0.8177 0.5765 0.9528 0.9528 0.9528 0.0985 0.3565

BEoM 0.3282 0.0010 0.0496 -0.0004 -0.0004 -0.0004 -0.0591 -0.0054 0.0000 0.9773 0.1658 0.9917 0.9917 0.9917 0.0989 0.8806

HS Ext -0.2151 -0.0532 -0.0584 0.0563 0.0563 0.0563 -0.0330 -0.0054 0.0000 0.1368 0.1028 0.1157 0.1157 0.1157 0.3564 0.8805 Upper-right cells and lower-left cells diagonal represent Pearson and Spearman correlation coefficients, respectively. P-Values in italic. Weekdays dummy variables not shown. 782 daily observations. 444

Table F.1 Pearson/Spearman correlation matrixes for coefficient estimates (continued)

Panel D: Quarterly observations MA SP500 SP500 Av Fed Fed Av GDP MKTC IP TD MA 0.3812 0.3371 0.6710 0.6604 -0.0553 0.2657 0.1032 0.0917 0.0027 0.0084 0.0000 0.0000 0.6749 0.0402 0.4325 0.4859

SP500 0.4612 0.9888 -0.1020 0.0209 0.7587 0.9593 0.8869 -0.0599 0.0004 0.0000 0.4380 0.8741 0.0000 0.0000 0.0000 0.6492

SP500 Av 0.4138 0.9827 -0.1276 -0.0055 0.7839 0.9566 0.9068 -0.0812 0.0015 0.0000 0.3314 0.9670 0.0000 0.0000 0.0000 0.5376

Fed 0.6267 -0.0492 -0.0600 0.9399 -0.5249 -0.2147 -0.4084 -0.0252 0.0000 0.7054 0.6449 0.0000 0.0000 0.0995 0.0012 0.8482

Fed Av 0.6054 0.0454 0.0402 0.8923 -0.4411 -0.1044 -0.3060 -0.0178 0.0000 0.7273 0.7576 0.0000 0.0004 0.4274 0.0174 0.8928

GDP -0.0565 0.6615 0.6896 -0.5597 -0.5160 0.8877 0.9597 -0.0213 0.6642 0.0000 0.0000 0.0000 0.0001 0.0000 0.0000 0.8719

MKTC 0.3238 0.9438 0.9420 -0.1996 -0.1275 0.8308 0.9480 -0.0489 0.0129 0.0000 0.0000 0.1253 0.3273 0.0000 0.0000 0.7108

IP 0.1206 0.8053 0.8313 -0.3675 -0.2990 0.9519 0.9270 -0.0438 0.3543 0.0000 0.0000 0.0048 0.0217 0.0000 0.0000 0.7395

TD -0.0141 -0.0833 -0.1100 -0.1175 -0.1009 0.0259 -0.0629 -0.0224 0.9134 0.5224 0.3983 0.3668 0.4382 0.8421 0.6292 0.8631 Upper-right cells and lower-left cells diagonal represent Pearson and Spearman correlation coefficients, respectively. P-Values in italic. Only non-dummy variables are exhibited. 60 quarterly observations.

445

Appendix G Regression Analysis for Hypotheses One and Two

Table G.1 Regression models outputs and tests results for hypothesis one

3112 2 j MAtjtiitlltmmtt=+αβ∑∑ Per + δ Month,,, + ∑ λ SP500 + ∑ ζ Fed + ω IP + ji==11 l = 1 m = 1 (19)

++++++φγϕθξεMKTCtttttt GDP TD Event MA −1

Backward elimination regression, with all variables left in the model significant at the 0.05 level: Panel A: Estimation and tests results for model using monthly M&A data T Standard Parameter Estimate P-Value Statistic Error Intercept -245.992 0.1057 -1.67637 146.741 Per -38.3226 0.0001 -4.68077 8.18724 Per_2 2.05267 0.0047 3.08964 0.66437 Per_3 -0.02809 0.0242 -2.3933 0.01173 Aug -58.0454 0.0039 -3.16688 18.3289 Nov -52.9105 0.0060 -2.99344 17.6755 SP500 0.35015 0.0003 4.23733 0.08263 Fed 32.7201 0.0297 2.30109 14.2194 TD 18.6442 0.0000 5.36609 3.47444 ANOVA F value 68.49 0.0000 R-squared 95.469% R2 Adjusted for d. f. 94.075% N used (read) 35 (36) Durbin-Watson D 2.07124 0.2912 Chi-Squared (13 d. f.) 4.77137 0.979887 Shapiro-Wilk W 0.95536 0.212621 Skewness Z-score 1.37276 0.169827 Kurtosis Z-score 1.10334 0.269877 Kolmogorov-Smirnov 0.081671 0.973699 Modified K-S D 0.494093 ≥ 0.10* Cramer-Von Mises W2 0.048242 0.52555* Watson U2 0.038493 0.64584* Anderson-Darling A2 0.386077 0.37226* Kuiper V 0.152129 ≥ 0.10* Box-Pierce Test 8.90604 0.630565 * P-Value has been compared to tables of critical values specially constructed for fitting the selected distribution. Except for the Chi-Squared Test, other P-values are based on general tables.

446

Table G.1 Regression models outputs and tests results for hypothesis one (continued)

34 2 2 j MAtjtiitlltmmt=+αβ∑∑ Per + δ Week,,, + ∑ λ SP500 + ∑ ζ Fed + ji==11 l = 1 m = 1 (20)

++++φϕθψξεE__ BoMtt TD Event t Event ED t ++ MA tt−1

Backward elimination regression, with all variables left in the model significant at the 0.05 level: Panel B: Estimation and tests results for model using weekly M&A data T Standard Parameter Estimate P-Value Statistic Error Intercept -28.3688 0.4957 -0.683 41.5356 Per -2.64935 0.0000 -4.93264 0.537105 Per_2 0.034402 0.0013 3.2878 0.010463 Per_3 -0.000118 0.0070 -2.73458 0.000043 Week1 13.6826 0.0033 2.98494 4.58386 Week3 7.85482 0.0259 2.25012 3.49085 E_BoM 22.113 0.0000 4.60669 4.80019 SP500 0.061473 0.0069 2.74118 0.022425 Fed 9.30303 0.0264 2.24266 4.14821 TD 15.7606 0.0000 5.9524 2.64778 ANOVA F value 49.60 0.0000 R-squared 75.480% R2 Adjusted for d. f. 73.958% N used (read) 155 (156) Durbin-Watson D 1.80846 0.1172 Chi-Squared (26 d. f.) 33.0323 0.161153 Shapiro-Wilk W 0.96908 0.037554 Skewness Z-score 1.8038 0.071262 Kurtosis Z-score 1.0161 0.309581 Kolmogorov-Smirnov 0.059553 0.64161 Modified K-S D 0.744899 ≥ 0.10* Cramer-Von Mises W2 0.078805 0.21315* Watson U2 0.059180 0.34597* Anderson-Darling A2 0.641829 0.09247* Kuiper V 0.096185 ≥ 0.10* Box-Pierce Test 23.7014 0.478771 * P-Value has been compared to tables of critical values specially constructed for fitting the selected distribution. Except for the Chi-Squared Test, other P-values are based on general tables.

447

Table G.1 Regression models outputs and tests results for hypothesis one (continued)

34 j MAtjtiitttt=+αβ∑∑ Per + δ Weekday, + λ SP500 + ζ Fed + ϕ Hol + ji==11 (21)

++++++ϖ HS__ Extttttttφθψξε E BoM Event Event _ ED MA −1

Backward elimination regression, with all variables left in the model significant at the 0.05 level: Panel C: Estimation and tests results for model using daily M&A data T Standard Parameter Estimate P-Value Statistic Error Intercept 13.1969 0.0221 2.288 5.76786 Per -0.04283 0.0000 -8.81313 0.004860 Per_2 0.00004 0.0000 7.65654 0.000005 Mon 8.74325 0.0000 11.3343 0.771398 Tue 6.03012 0.0000 7.8878 0.764487 Wed 4.21136 0.0000 5.5252 0.762209 Thu 3.79882 0.0000 5.00339 0.759248 Hol -17.5629 0.0000 -12.8825 1.36332 HS_Ext -11.9676 0.0000 -6.72071 1.7807 E_BoM 12.6377 0.0000 15.1327 0.835129 SP500 0.01008 0.0086 2.62597 0.003841 MA_lag 0.09517 0.0009 3.33449 0.028543 ANOVA F value 95.34 0.0000 R-squared 57.694% R2 Adjusted for d. f. 57.088% N used (read) 781 (782) Durbin-Watson D 2.00652 0.4637 Chi-Squared (52 d. f.) 41.3188 0.855999 Shapiro-Wilk W 0.98240 0.077732 Skewness Z-score 3.37554 0.000736 Kurtosis Z-score 4.67686 0.000002 Kolmogorov-Smirnov 0.02704 0.617458 Modified K-S D 0.75912 ≥ 0.10 Cramer-Von Mises W2 0.16988 ≥ 0.10 Watson U2 0.13283 ≥ 0.10 Anderson-Darling A2 1.2108 ≥ 0.10 Kuiper V 0.04792 ≥ 0.10 Box-Pierce Test 36.9366 0.044401

448

Table G.1 Regression models outputs and tests results for hypothesis one (continued)

322 Log MAtiit=+αδ∑∑∑Quarter ,, + λllt SP500 , + ζ mmtt Fed + ω IP + ilm===111 (18)

+++++φ MKTCttttγϕθξ GDP TD Event Log MAt−1 + ε t

Backward elimination regression, with all variables left in the model significant at the 0.05 level: Panel D: Estimation and tests results for model using quarterly M&A data T Standard Parameter Estimate P-Value Statistic Error Intercept 1.22817 0.0000 4.45549 0.275653 SP500 -0.000253 0.0012 -3.43246 0.000073 GDP -0.000062 0.0000 -5.25043 0.000011 MKTC 0.000032 0.0000 4.50368 0.000007 Log MA_lag 0.76621 0.0000 10.3664 0.073912 ANOVA F value 66.13 0.0000 R-squared 83.04% R2 Adjusted for d. f. 81.78% N used (read) 59 (60) Durbin-Watson D 2.00502 0.3195 Chi-Squared (17 d. f.) 15.2375 0.578382 Shapiro-Wilk W 0.97507 0.483264 Skewness Z-score 0.32828 0.742692 Kurtosis Z-score 0.62073 0.534775 Kolmogorov-Smirnov 0.08985 0.72748 Modified K-S D 0.69924 ≥ 0.10* Cramer-Von Mises W2 0.05490 0.43549* Watson U2 0.05457 0.39812* Anderson-Darling A2 0.36163 0.433559 Kuiper V 0.14359 ≥ 0.10* Box-Pierce Test 24.2229 0.187742 * P-Value has been compared to tables of critical values specially constructed for fitting the selected distribution. Except for the Chi-Squared Test, other P-values are based on general tables.

449

Table G.2 Regression models outputs and tests results for hypothesis two

3112 2 j WITHtjtiitmmt=+αβ∑∑ Per + δ Month,, + ∑ λllt SP500 , + ∑ ζ Fed + ji==11 l = 1 m = 1 (23) +++++++φγωϕρψMKTCttttt GDP IP TD MA Event t

++ξεWITHt−1 t Backward elimination regression, with all variables left in the model significant at the 0.05 level: Panel A: Estimation and tests results for model using monthly WITH data T Standard Parameter Estimate P-Value Statistic Error Intercept -46.4071 0.0285 -2.32547 19.956 Per -9.00213 0.0000 -6.27822 1.43387 Per_2 0.57598 0.0000 5.68897 0.10124 Per_3 -0.00916 0.0000 -5.39809 0.00169 Aug -9.58472 0.0025 -3.36588 2.84761 Oct -6.79894 0.0133 -2.66549 2.55073 Fed 9.80755 0.0001 4.60347 2.13047 MKTC 0.00409 0.0012 3.66945 0.00111 MA -0.07593 0.0034 -3.23117 0.02349 TD 1.85311 0.0117 2.71929 0.68146 ANOVA F value 20.80 0.0000 R-squared 88.218% R2 Adjusted for d. f. 83.976% N used (read) 35 (36) Durbin-Watson D 2.28273 0.5118 Chi-Squared (13 d. f.) 16.6575 0.21544 Shapiro-Wilk W 0.94590 0.10997 Skewness Z-score 0.96900 0.33254 Kurtosis Z-score -0.61912 0.53583 Kolmogorov-Smirnov 0.15148 0.40288 Modified K-S D 0.91642 <0.05* (≥0.10) Cramer-Von Mises W2 0.11419 0.0683* Watson U2 0.10532 0.0710* Anderson-Darling A2 0.63708 0.0889* Kuiper V 0.22084 ≥0.10* Box-Pierce Test 6.46306 0.84075 * P-Value has been compared to tables of critical values specially constructed for fitting the selected distribution. Except for the Chi-Squared Test, other P-values are based on general tables.

450

Table G.2 Regression models outputs and tests results for hypothesis two (continued)

322 WITHtiit=+αδ∑∑∑ Quarter,, + λllt SP500 , + ζ mmt Fed +φ MKTCt + ilm===111 (24)

+++++γ GDPttωϕ IP TD t ρ MA t ψ Event t + ξ WITHt−1 + ε t

Backward elimination regression, with all variables left in the model significant at the 0.05 level: Panel B: Estimation and tests results for model using quarterly WITH data T Standard Parameter Estimate P-Value Statistic Error Intercept 61.7643 0.5862 0.54767 112.776 SP500 Av 0.12880 0.0000 6.00736 0.02144 Fed 5.02267 0.0007 3.61398 1.38979 MKTC -0.00652 0.0015 -3.35554 0.00194 IP 61.7643 0.0002 0.54767 112.776 TD 0.12880 0.0126 6.00736 0.02144 ANOVA F value 65.03 0.0000 R-squared 85.98% R2 Adjusted for d. f. 84.66% N used (read) 59 (60) Durbin-Watson D 1.9579 0.2741 Chi-Squared (17 d. f.) 14.5593 0.62719 Shapiro-Wilk W 0.98238 0.77596 Skewness Z-score 0.97291 0.33059 Kurtosis Z-score 0.64756 0.51726 Kolmogorov-Smirnov 0.05177 0.99741 Modified K-S D 0.40291 ≥0.10* Cramer-Von Mises W2 0.02990 0.8453* Watson U2 0.02498 0.8920* Anderson-Darling A2 0.21435 0.8427* Kuiper V 0.09383 ≥0.10* Box-Pierce Test 16.4665 0.62596 * P-Value has been compared to tables of critical values specially constructed for fitting the selected distribution. Except for the Chi-Squared Test, other P-values are based on general tables.

451

Table G.3 Résumé of regression analysis for hypothesis one and two Selected Hypothesis One Hypothesis Two Variables† Quarterly Monthly Weekly Daily Quarterly Monthly

Intercept 1.22817 * -245.992 -28.3688 13.1969 ** 61.7643 -46.4071 ** Per - -38.3226 * -2.64935 * -0.04283 * - -9.00213 * Per_2 - 2.05267 * 0.03440 * 0.00004 * - 0.57598 * Per_3 - -0.02809 ** -0.00011 * - - -0.00916 * SP500 -0.00025 * 0.35015 * 0.06147 * 0.01008 * - -

SP500_Av - - - - 0.12880 * - Fed - 32.7201 ** 9.30303 ** - 5.02267 * 9.80755 * GDP -0.00006 * - - - - - MKTC 0.000032 * - - - -0.0065 * 0.00409 * IP - - - - 61.7643 * - Aug - -58.0454 * - - - -9.58472 * Oct ------6.79894 ** Nov - -52.9105 * - - - - Week1 - - 13.6826 * - - - Week3 - - 7.85482 ** - - - Mon - - - 8.74325 * - - Tue - - - 6.03012 * - - Wed - - - 4.21136 * - - Thu - - - 3.79882 * - - TD - 18.6442 * 15.7606 * - 0.12880 ** 1.85311 ** E_BoM - - 22.1130 * 12.6377 * - - Hol - - - -17.5629 * - - HS_Ext - - - -11.9676 * - - MA ------0.07593 * MA_lag 0.76621 *‡ - - 0.09517 * - - F value 66.13 * 68.49 * 49.60 * 95.34 * 65.03 * 20.80 * Adj. R2 81.78% 94.08% 73.96% 57.09% 84.66% 83.97% N read 60 36 156 782 60 36 * Significant at 0.01 level ** Significant at 0.05 level *** Significant at 0.10 level † The models have employed adjusted sets of variables according to their nature, being worthwhile to recall that Event_ED variable was not used in the models employing quarterly and monthly data. ‡ Log MA_lag.

452

Appendix H Residuals’ Autocorrelation Tables and Correlograms

Table H.1 Autocorrelations for hypothesis one

Panel A: Estimated autocorrelations for residuals of model using monthly data Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 -0.052813 0.169031 -0.331295 0.331295 2 -0.282255 0.169502 -0.332218 0.332218 3 -0.254428 0.182437 -0.357571 0.357571 4 -0.056948 0.192308 -0.376917 0.376917 5 0.113448 0.192789 -0.37786 0.37786 6 -0.000987 0.194687 -0.381581 0.381581 7 0.146001 0.194687 -0.381581 0.381581 8 -0.1558 0.197791 -0.387664 0.387664 9 -0.087075 0.201267 -0.394476 0.394476 10 0.094923 0.20234 -0.39658 0.39658 11 0.170207 0.203609 -0.399066 0.399066

453

Table H.1 Autocorrelations for hypothesis one (continued)

Panel B: Estimated autocorrelations for residuals of model using weekly data Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 0.092073 0.0803219 -0.157428 0.157428 2 -0.071949 0.081 -0.158757 0.158757 3 -0.12655 0.0814113 -0.159563 0.159563 4 -0.120454 0.0826707 -0.162032 0.162032 5 -0.015722 0.0837953 -0.164236 0.164236 6 -0.144914 0.0838143 -0.164273 0.164273 7 -0.057355 0.0854155 -0.167412 0.167412 8 0.038933 0.0856636 -0.167898 0.167898 9 0.1028 0.0857777 -0.168122 0.168122 10 0.114209 0.0865689 -0.169672 0.169672 11 -0.011530 0.0875356 -0.171567 0.171567 12 -0.104217 0.0875454 -0.171586 0.171586 13 -0.086147 0.0883422 -0.173148 0.173148 14 0.051623 0.0888825 -0.174207 0.174207 15 0.011965 0.0890757 -0.174586 0.174586 16 -0.072273 0.0890861 -0.174606 0.174606 17 -0.120016 0.0894636 -0.175346 0.175346 18 -0.055463 0.0904964 -0.17737 0.17737 19 0.013334 0.0907154 -0.177799 0.177799 20 0.083738 0.090728 -0.177824 0.177824 21 -0.023965 0.0912253 -0.178799 0.178799 22 -0.046902 0.0912659 -0.178878 0.178878 23 0.067750 0.0914213 -0.179183 0.179183 24 0.024266 0.0917446 -0.179817 0.179817

454

Table H.1 Autocorrelations for hypothesis one (continued)

Panel C: Estimated autocorrelations for residuals of model using daily data Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limi† Prob. Limit† 1 -0.0048219 0.0357828 -0.0701332 0.0701332 2 0.0076419 0.0357836 -0.0701348 0.0701348 3 0.0574489 0.0357857 -0.0701389 0.0701389 4 0.0884542 0.0359036 -0.07037 0.07037 5 0.0055488 0.0361816 -0.0709147 0.0709147 6 0.0135122 0.0361827 -0.0709169 0.0709169 7 0.047677 0.0361891 -0.0709295 0.0709295 8 -0.015261 0.0362695 -0.071087 0.071087 9 -0.0576198 0.0362777 -0.0711031 0.0711031 10 0.0328195 0.0363947 -0.0713324 0.0713324 11 0.0715474 0.0364326 -0.0714066 0.0714066 12 -0.0877919 0.036612 -0.0717584 0.0717584 13 -0.0676372 0.0368806 -0.0722847 0.0722847 14 -0.0411988 0.0370391 -0.0725954 0.0725954 15 0.0101187 0.0370977 -0.0727103 0.0727103 16 -0.0156709 0.0371012 -0.0727172 0.0727172 17 -0.0512292 0.0371097 -0.0727338 0.0727338 18 -0.010793 0.0372001 -0.0729111 0.0729111 19 -0.0390522 0.0372042 -0.0729189 0.0729189 20 0.0089166 0.0372566 -0.0730217 0.0730217 21 0.0102154 0.0372593 -0.0730271 0.0730271 22 0.0466923 0.0372629 -0.0730341 0.0730341 23 -0.0401955 0.0373378 -0.0731808 0.0731808 24 -0.0357627 0.0373931 -0.0732893 0.0732893 Values outside autocorrelations limits at the 95% confidence level in italic. † Autocorrelations limits at the 99% confidence level: ±0.0924818 at fourth lag, ±0.0938442 at eleventh lag, and ±0.0943065 at twelfth lag.

455

Table H.1 Autocorrelations for hypothesis one (continued)

Panel D: Estimated autocorrelations for residuals of model using quarterly data Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 -0.022356 0.130189 -0.255166 0.255166 2 0.162279 0.130254 -0.255294 0.255294 3 0.204205 0.133637 -0.261924 0.261924 4 0.179265 0.138825 -0.272092 0.272092 5 -0.000274 0.142694 -0.279676 0.279676 6 -0.12747 0.142694 -0.279677 0.279677 7 -0.048420 0.144612 -0.283434 0.283434 8 -0.012906 0.144886 -0.283972 0.283972 9 -0.185057 0.144906 -0.28401 0.28401 10 -0.263525 0.148857 -0.291756 0.291756 11 -0.108787 0.156565 -0.306862 0.306862 12 -0.191542 0.157841 -0.309363 0.309363 13 -0.155496 0.161733 -0.316991 0.316991 14 -0.214075 0.164247 -0.321919 0.321919 15 -0.185775 0.16891 -0.331058 0.331058 16 -0.016555 0.172338 -0.337777 0.337777 17 -0.039167 0.172365 -0.33783 0.33783 18 -0.140723 0.172516 -0.338126 0.338126 19 0.113057 0.174451 -0.341918 0.341918

456

Table H.2 Autocorrelations for hypothesis two

Panel A: Estimated autocorrelations for residuals of monthly data model Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 -0.18472 0.169031 -0.331295 0.331295 2 -0.076701 0.174703 -0.342413 0.342413 3 -0.022913 0.175663 -0.344293 0.344293 4 -0.186152 0.175748 -0.344461 0.344461 5 -0.035102 0.181294 -0.355331 0.355331 6 0.008586 0.181488 -0.355711 0.355711 7 -0.096937 0.1815 -0.355734 0.355734 8 -0.10906 0.182973 -0.358621 0.358621 9 0.127963 0.184821 -0.362243 0.362243 10 0.23195 0.187335 -0.367171 0.367171 11 -0.129246 0.195368 -0.382916 0.382916

457

Fig. H.2 Autocorrelations for hypothesis two (continued)

Panel B: Estimated autocorrelations for residuals of quarterly data model Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 0.009070 0.130189 -0.255166 0.255166 2 0.145194 0.1302 -0.255187 0.255187 3 -0.19348 0.132916 -0.26051 0.26051 4 -0.022855 0.137606 -0.269704 0.269704 5 -0.175231 0.137671 -0.26983 0.26983 6 0.066268 0.141401 -0.27714 0.27714 7 0.095012 0.141926 -0.27817 0.27817 8 0.208951 0.143 -0.280275 0.280275 9 0.030555 0.148084 -0.290241 0.290241 10 -0.209988 0.148191 -0.29045 0.29045 11 -0.084516 0.153152 -0.300172 0.300172 12 -0.205582 0.15394 -0.301717 0.301717 13 -0.118073 0.158525 -0.310704 0.310704 14 -0.035870 0.160009 -0.313612 0.313612 15 0.077841 0.160145 -0.313879 0.313879 16 0.103329 0.160785 -0.315133 0.315133 17 0.026558 0.161907 -0.317332 0.317332 18 -0.057702 0.16198 -0.317476 0.317476 19 -0.041870 0.162328 -0.318158 0.318158

458

Table H.3 Partial autocorrelations for hypothesis one

Panel A: Estimated partial autocorrelations for model using monthly data Partial Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 -0.052813 0.169031 -0.331295 0.331295 2 -0.285842 0.169031 -0.331295 0.331295 3 -0.315257 0.169031 -0.331295 0.331295 4 -0.243145 0.169031 -0.331295 0.331295 5 -0.141458 0.169031 -0.331295 0.331295 6 -0.242315 0.169031 -0.331295 0.331295 7 0.007533 0.169031 -0.331295 0.331295 8 -0.273265 0.169031 -0.331295 0.331295 9 -0.223778 0.169031 -0.331295 0.331295 10 -0.111447 0.169031 -0.331295 0.331295 11 0.005124 0.169031 -0.331295 0.331295

459

Table H.3 Partial autocorrelations for hypothesis one (continued)

Panel B: Estimated partial autocorrelations for model using weekly data Partial Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit† Prob. Limit† 1 0.0920731 0.0803219 -0.157428 0.157428 2 -0.0811147 0.0803219 -0.157428 0.157428 3 -0.113624 0.0803219 -0.157428 0.157428 4 -0.106537 0.0803219 -0.157428 0.157428 5 -0.0143037 0.0803219 -0.157428 0.157428 6 -0.179762 0.0803219 -0.157428 0.157428 7 -0.0663485 0.0803219 -0.157428 0.157428 8 0.0037977 0.0803219 -0.157428 0.157428 9 0.0493395 0.0803219 -0.157428 0.157428 10 0.0578553 0.0803219 -0.157428 0.157428 11 -0.0229596 0.0803219 -0.157428 0.157428 12 -0.095774 0.0803219 -0.157428 0.157428 13 -0.0615547 0.0803219 -0.157428 0.157428 14 0.070004 0.0803219 -0.157428 0.157428 15 -0.0085935 0.0803219 -0.157428 0.157428 16 -0.0757551 0.0803219 -0.157428 0.157428 17 -0.125864 0.0803219 -0.157428 0.157428 18 -0.0894381 0.0803219 -0.157428 0.157428 19 -0.0657625 0.0803219 -0.157428 0.157428 20 0.0434247 0.0803219 -0.157428 0.157428 21 -0.0639771 0.0803219 -0.157428 0.157428 22 -0.0724003 0.0803219 -0.157428 0.157428 23 0.0269906 0.0803219 -0.157428 0.157428 24 -0.0430147 0.0803219 -0.157428 0.157428 Values outside autocorrelations limits at the 95% confidence level in italic. † Autocorrelations limits at the 99% confidence level: ±0.206896.

460

Table H.3 Partial autocorrelations for hypothesis one (continued)

Panel C: Estimated partial autocorrelations for model using daily data Partial Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit† Prob. Limit† 1 -0.0048219 0.0357828 -0.0701332 0.0701332 2 0.0076188 0.0357828 -0.0701332 0.0701332 3 0.0575269 0.0357828 -0.0701332 0.0701332 4 0.089274 0.0357828 -0.0701332 0.0701332 5 0.0061086 0.0357828 -0.0701332 0.0701332 6 0.0090890 0.0357828 -0.0701332 0.0701332 7 0.0380153 0.0357828 -0.0701332 0.0701332 8 -0.023337 0.0357828 -0.0701332 0.0701332 9 -0.0617151 0.0357828 -0.0701332 0.0701332 10 0.0253541 0.0357828 -0.0701332 0.0701332 11 0.0686951 0.0357828 -0.0701332 0.0701332 12 -0.0793218 0.0357828 -0.0701332 0.0701332 13 -0.0659414 0.0357828 -0.0701332 0.0701332 14 -0.054861 0.0357828 -0.0701332 0.0701332 15 0.0106493 0.0357828 -0.0701332 0.0701332 16 0.0100143 0.0357828 -0.0701332 0.0701332 17 -0.0413918 0.0357828 -0.0701332 0.0701332 18 -0.0098634 0.0357828 -0.0701332 0.0701332 19 -0.0247946 0.0357828 -0.0701332 0.0701332 20 0.0251668 0.0357828 -0.0701332 0.0701332 21 0.0080918 0.0357828 -0.0701332 0.0701332 22 0.0447019 0.0357828 -0.0701332 0.0701332 23 -0.0238114 0.0357828 -0.0701332 0.0701332 24 -0.0320228 0.0357828 -0.0701332 0.0701332 Values outside autocorrelations limits at the 95% confidence level in italic. † Autocorrelations limits at the 99% confidence level: ±0.0921706.

461

Table H.3 Partial autocorrelations for hypothesis one (continued)

Panel D: Estimated partial autocorrelations for model using quarterly data Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit† Prob. Limit† 1 -0.022356 0.130189 -0.255166 0.255166 2 0.16186 0.130189 -0.255166 0.255166 3 0.216646 0.130189 -0.255166 0.255166 4 0.181291 0.130189 -0.255166 0.255166 5 -0.049060 0.130189 -0.255166 0.255166 6 -0.258334 0.130189 -0.255166 0.255166 7 -0.181277 0.130189 -0.255166 0.255166 8 0.003929 0.130189 -0.255166 0.255166 9 -0.053409 0.130189 -0.255166 0.255166 10 -0.195992 0.130189 -0.255166 0.255166 11 -0.109621 0.130189 -0.255166 0.255166 12 -0.140429 0.130189 -0.255166 0.255166 13 -0.037179 0.130189 -0.255166 0.255166 14 -0.082031 0.130189 -0.255166 0.255166 15 -0.195368 0.130189 -0.255166 0.255166 16 -0.060235 0.130189 -0.255166 0.255166 17 0.038286 0.130189 -0.255166 0.255166 18 -0.111679 0.130189 -0.255166 0.255166 19 0.020714 0.130189 -0.255166 0.255166 Value outside autocorrelations limits at the 95% confidence level in italic. † Autocorrelations limits at the 99% confidence level: ±0.335345.

462

Table H.4 Partial autocorrelations for hypothesis two

Panel A: Estimated autocorrelations for residuals of monthly data model Partial Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 -0.18472 0.169031 -0.3313 0.331295 2 -0.114738 0.169031 -0.3313 0.331295 3 -0.062845 0.169031 -0.3313 0.331295 4 -0.22376 0.169031 -0.3313 0.331295 5 -0.145017 0.169031 -0.3313 0.331295 6 -0.093019 0.169031 -0.3313 0.331295 7 -0.1885 0.169031 -0.3313 0.331295 8 -0.29461 0.169031 -0.3313 0.331295 9 -0.099792 0.169031 -0.3313 0.331295 10 0.155445 0.169031 -0.3313 0.331295 11 -0.146222 0.169031 -0.3313 0.331295

463

Table H.4 Partial autocorrelations for hypothesis two (continued)

Panel B: Estimated autocorrelations for residuals of quarterly data model Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit† Prob. Limit† 1 0.009070 0.130189 -0.255166 0.255166 2 0.145123 0.130189 -0.255166 0.255166 3 -0.200154 0.130189 -0.255166 0.255166 4 -0.037477 0.130189 -0.255166 0.255166 5 -0.123537 0.130189 -0.255166 0.255166 6 0.048401 0.130189 -0.255166 0.255166 7 0.131955 0.130189 -0.255166 0.255166 8 0.14621 0.130189 -0.255166 0.255166 9 0.005764 0.130189 -0.255166 0.255166 10 -0.271793 0.130189 -0.255166 0.255166 11 -0.015166 0.130189 -0.255166 0.255166 12 -0.108887 0.130189 -0.255166 0.255166 13 -0.147666 0.130189 -0.255166 0.255166 14 -0.031987 0.130189 -0.255166 0.255166 15 -0.045914 0.130189 -0.255166 0.255166 16 0.059521 0.130189 -0.255166 0.255166 17 0.010426 0.130189 -0.255166 0.255166 18 -0.002216 0.130189 -0.255166 0.255166 19 0.032174 0.130189 -0.255166 0.255166 Value outside autocorrelations limits at the 95% confidence level in italic. † Autocorrelations limits at the 99% confidence level: ±0.335345.

464

Fig. H.1 Autocorrelations correlograms for hypothesis one

Panel A: Model using monthly data

Estimated Autocorrelations for RESIDUALS

0.6

0.2

-0.2 Autocorrelations -0.6

-1 024681012 lag

Panel B: Model using weekly data

Estimated Autocorrelations for RESIDUALS

0.6

0.2

-0.2 Autocorrelations -0.6

-1 0 5 10 15 20 25 lag

Panel C: Model using daily data

Estimated Autocorrelations for RESIDUALS

0.6

0.2

-0.2 Autocorrelations -0.6

-1 0 5 10 15 20 25 lag

465

Fig. H.1 Autocorrelations correlograms for hypothesis one (continued)

Panel D: Model using quarterly data

Estimated Autocorrelations for RESIDUALS

0.6

0.2

-0.2 Autocorrelations -0.6

-1 0 4 8 12 16 20 lag

Fig. H.2 Autocorrelations correlogram for hypothesis two

Panel A: Model using monthly data

Estimated Autocorrelations for RESIDUALS

0.6

0.2

-0.2 Autocorrelations -0.6

-1 024681012 lag

Panel B: Model using quarterly data

Estimated Autocorrelations for RESIDUALS

0.6

0.2

-0.2 Autocorrelations -0.6

-1 0 4 8 12 16 20 lag

466

Appendix I Durbin-Watson Distribution and Critical Values

Fig. I.1 The five regions of the Durbin-Watson d-statistic

Reject HO Inconclusive Accept HO Inconclusive Reject HO

Significant Uncertainty Not Uncertainty Significant Positive Zone Significant Zone Negative Autocorrelation Autocorrelation Autocorrelation

0 DWL DWU 2 4-DWU 4-DWL 4

DWL for Durbin-Watson lower critical value.

DWU for Durbin-Watson upper critical value.

467

Table I.1 Critical Values for the Durbin-Watson test

a n k DWL DWU 4 - DWU 4 - DWL 35 9 0.97099 2.05436 1.94564 3.02901 35 10 0.90788 2.14395 1.85605 3.09212 59 5 1.43848 1.72663 2.27337 2.56152 59 6 1.40191 1.76720 2.2328 2.59809 155 10 1.61643 1.86142 2.13858 2.38357 740 10 1.856† 1.904† 2.096† 2.144† 781 12 1.855† 1.911† 2.089† 2.145† Table entries for 5% significance level, one-sided test of the Durbin-Watson statistic. a k includes intercept (k-1 for number of explanatory variables).

† Approximate value (own estimation). Sources: Savin & White (1977); Website of Stanford University (Online) (2007).

468

Appendix J Plots for Normal Distribution of Residuals

Fig. J.1 Plot of residuals for hypothesis one

Panel A: Model using monthly data

Residual Plot

0 residual

-40

-80 0 10203040 row number

Panel B: Model using weekly data

Residual Plot

0 residual -20

-40

-60 0 40 80 120 160 row number

Panel C: Model using daily data

Residual Plot

0 residual -10

-20

-30 0 200 400 600 800 row number

469

Fig. J.1 Plot of residuals for hypothesis one (continued)

Panel D: Model using quarterly data

Resi dual Pl ot

0.1

0.06

0.02

residual -0.02

-0.06

-0.1 0 102030405060 row number

Fig. J.2 Plot of residuals for hypothesis two

Panel A: Model using monthly data

Residual Plot

0 residual

-4

-8 0 10203040 row number

Panel B: Model using quarterly data

Resi dual Pl ot

1 residual

-19

-39 0 102030405060 row number

470

Fig. J.3 Histograms of residuals with normal distribution curve superimposed for hypothesis one

Panel A: Model using monthly data

Histogram f or RESIDUALS

15 Distribution Normal 12

6 frequency

0 -80 -50 -20 10 40 70 RESI DUALS

Panel B: Model using weekly data

Histogram f or RESIDUALS

50 Distribution Normal 40

20 frequency

0 -50 -30 -10 10 30 50 70 RESI DUALS

Panel C: Model using daily data

Histogram f or RESIDUALS

300 Distribution Normal 250

200

150

frequency 100

0 -26 -16 -6 4 14 24 34 RESI DUALS

471

Fig. J.3 Histograms of residuals with normal distribution curve superimposed for hypothesis one

Panel D: Model using quarterly data

Histogram f or RESIDUALS

24 Distribution Normal 20

frequency 8

0 -0.11 -0.07 -0.03 0.01 0.05 0.09 RESI DUALS

Fig. J.4 Histogram of residuals with normal distribution curve superimposed for hypothesis two

Panel A: Model using monthly data

Histogram f or RESIDUALS

10 Distribution Normal 8

4 frequency

0 -6 -3 0 3 6 9 RESI DUALS

Panel B: Model using quarterly data

Histogram f or RESIDUALS

24 Distribution Normal 20

frequency 8

0 -43 -23 -3 17 37 RESI DUALS

472

Fig. J.5 Residuals’ normal probability plots for hypothesis one

Panel A: Model using monthly data

Quantile-Quantile Plot

70 Distribution Normal 40

-20 RESIDUALS -50

-80 -80 -50 -20 10 40 70 Normal distribution

Panel B: Model using weekly data

Quantile-Quantile Plot

55 Distribution Normal 35

-5 RESIDUALS -25

-45 -45 -25 -5 15 35 55 Normal distribution

Panel C: Model using daily data

Quantile-Quantile Plot

36 Distribution Normal 26

-4 RESIDUALS -14

-24 -24 -14 -4 6 16 26 36 Normal distribution

473

Fig. J.5 Residuals’ normal probability plots for hypothesis one (continued)

Panel D: Model using quarterly data

Quanti l e- Quanti l e Pl ot

0.08 Di str i buti on Nor mal 0.05

0.02

-0.01

-0.04 RESIDUALS

-0.07

-0.1 -0.1 -0.07 -0.04 -0.01 0.02 0.05 0.08 Nor mal di str i buti on

Fig. J.6 Residuals’ normal probability plot for hypothesis two

Panel A: Model using monthly data

Quantile-Quantile Plot

8 Distribution Nor mal 5

-1 RESIDUALS -4

-7 -7 -4 -1 2 5 8 Normal distribution

Panel B: Model using quarterly data

Quanti l e- Quanti l e Pl ot

41 Di str i buti on Nor mal

1 RESIDUALS -19

-39 -39 -19 1 21 41 Nor mal di str i buti on

474

Appendix K Outliers and Influential Points

Table K.1 Unusual residuals for hypothesis one

Panel A: Model using monthly data Studentised Studentised Modified Studentised Values Without Values With MAD Row Residual Residual Deletion Deletion Z-Score 4 -71.8512 -3.49 -2.92824 -3.4513 -2.504 15 -48.7819 -2.16 -1.98807 -2.14889 -1.70519 Excessive studentised residual in italic.

Panel B: Model using weekly data Studentised Studentised Modified Studentised Values Without Values With MAD Row Residual Residual Deletion Deletion Z-Score 5 39.5832 2.46 2.35398 2.40552 2.39801 26 51.0338 3.08 3.03493 3.14071 3.07918 79 39.6534 2.43 2.35815 2.40994 2.40218 103 43.9561 2.67 2.61403 2.68304 2.65814 126 35.3771 2.14 2.10384 2.14182 2.1478 135 35.3992 2.15 2.10516 2.1432 2.14911 139 -34.2884 -2.05 -2.0391 -2.07404 -1.99646 153 -36.8139 -2.32 -2.18929 -2.23157 -2.1467 155 50.6165 3.27 3.01012 3.1134 3.05436 Excessive studentised residuals in italic.

475

Table K.1 Unusual residuals for hypothesis one (continued)

Panel C: Model using daily data Studentised Studentised Modified Studentised Values Without Values With MAD Row Residual Residual Deletion Deletion Z-Score 21 28.2386 4.34 4.24816 4.30099 4.55937 37 26.2322 4.01 3.94632 3.98892 4.23865 65 20.6055 3.13 3.09985 3.12115 3.33921 66 -23.0304 -3.54 -3.46465 -3.49389 -3.63608 130 29.4753 4.51 4.43421 4.49413 4.75706 586 27.5797 4.21 4.14904 4.19834 4.45405 Excessive studentised residuals and modified MAD Z-scores in italic.

Panel D: Model using quarterly data Studentised Studentised Modified Studentised Values Without Values With MAD Row Residual Residual Deletion Deletion Z-Score 31 -0.09861 -3.26 -3.04919 -3.36114 -3.19579 Excessive studentised residual in italic.

476

Table K.2 Unusual residuals for hypothesis two

Panel A: Model using monthly data Studentised Studentised Modified Studentised Values Without Values With MAD Row Residual Residual Deletion Deletion Z-Score 26 7.52585 2.48 2.40341 2.68336 2.37721 36 5.13396 2.29 1.63955 1.73486 1.70375 There are no excessive studentised residuals or modified MAD Z-scores.

Panel B: Model using quarterly data Studentised Studentised Modified Studentized Values Without Values With MAD Row Residual Residual Deletion Deletion Z-Score 23 -38.9292 -3.35 -2.96044 -3.24522 -3.02038 47 -28.8337 -2.24 -2.19271 -2.31078 -2.2569 57 26.3563 2.09 2.00431 2.09642 1.91689 Excessive studentised residual in italic.

477

Fig. K.1 Box-and-Whisker plot of residuals for hypothesis one

Panel A: Model using monthly data

-80-60-40-200 20406080 RESIDUALS

Panel B: Model using weekly data

-45 -30 -15 0 15 30 45 RESIDUALS

Panel C: Model using daily data

-30 -20 -10 0 10 20 30 RESIDUALS

478

Box-and-Whisker plot of residuals for hypothesis one (Continued)

Panel D: Model using quarterly data

-0.1 -0.07 -0.04 -0.01 0.02 0.05 0.08 RESIDUALS

Fig. K.2 Box-and-Whisker plot of residuals for hypothesis two

Panel A: Model using monthly data

-9-6-30369 RESIDUALS

Panel B: Model using quarterly data

-39 -19 1 21 41 RESIDUALS

479

Fig. K.3 Plot of outliers for hypothesis one

Panel A: Model using monthly data

Outlier Plot with Sigma Limits Sample mean = 0.00000462857, std. deviation = 24.5373 100 4 3 60 2

20 1 0 -20 -1 RESIDUALS -2 -60 -3 -100 -4 010203040 Row number

Panel B: Model using weekly data

Outlier Plot with Sigma Limits Sample mean = 0.00000287355, std. deviation = 16.8155 80 4 50 3 2 20 1 0 -10 -1 RESIDUALS -2 -40 -3 -70 -4 04080120160 Row number

480

Fig. K.3 Plot of outliers for hypothesis one (continued)

Panel C: Model using daily data

Outlier Plot with Sigma Limits Sample mean = 2.10986E-7, std. deviation = 6.64725 33 4 23 3 13 2 1 3 0 -1

RESIDUALS -7 -2 -17 -3 -27 -4 0 200 400 600 800 Row number

Panel D: Model using quarterly data

Outlier Plot with Sigma Limits Sample mean = -1.35593E-9, std. deviation = 0.0323609 0.17

0.12 4 3 0.07 2 1 0.02 0

RESIDUALS -0.03 -1 -2 -0.08 -3 -0.13 -4 0 102030405060 Row number

481

Fig. K.4 Plot of outliers for hypothesis two

Panel A: Model using monthly data

Outlier Plot with Sigma Limits Sample mean = -2.E-8, std. deviation = 3.13132 17

12 4 3 7 2 1 2 0

RESIDUALS -3 -1 -2 -8 -3 -13 -4 010203040 Row number

Panel B: Model using quarterly data

Outlier Plot with Sigma Limits Sample mean = 0.00000183051, std. deviation = 13.1498 60 4 40 3 2 20 1 0 0 -1

RESIDUALS -20 -2 -40 -3 -4 -60 0 102030405060 Row number

482

Table K.3 Influential points for hypothesis one

Panel A: Model using monthly data Mahalanobis Row Leverage Distance DFITS COOK’s D 2 0.478847 29.3505 1.14009 0.1421550 4 0.230409 8.90934 -1.91005 0.2834294 11 0.517912 34.4816 -1.10874 0.1358353 15 0.259749 10.6089 -1.27837 0.1591881 20 0.421596 23.0829 -1.1974 0.1535954 32 0.429162 23.8391 1.32439 0.1853844 Average leverage of single data point = 0.257143. Excessive Leverage, DFITS, and Cook’s D values in italic.

Panel B: Model using weekly data Mahalanobis Row Leverage Distance DFITS COOK’s D 5 0.109032 17.7298 0.86116 0.0716604 26 0.034742 4.51349 0.58512 0.0323388 79 0.081673 12.6139 0.72401 0.0507079 89 0.351172 81.8164 0.47447 0.0226034 103 0.057117 8.27487 0.65633 0.0413354 104 0.084174 13.0688 -0.53353 0.0280598 126 0.064021 9.4718 0.55860 0.0304556 135 0.077157 11.7986 0.62250 0.0378041 153 0.134014 22.6837 -0.91145 0.0806455 155 0.146825 25.3366 1.35482 0.1720823 Average leverage of single data point = 0.0645161. Excessive Leverage, DFITS, and Cook’s D values in italic.

483

Table K.3 Influential points for hypothesis one (continued)

Panel C: Model using daily data Mahalanobis Row Leverage Distance DFITS COOK’s D 6 0.030427 23.4481 0.28761 0.0068790 7 0.026583 20.2752 -0.33444 0.0092836 11 0.050007 40.0081 0.35620 0.0105543 21 0.031430 24.2803 0.78096 0.0496760 22 0.054978 44.3212 0.27497 0.0062986 35 0.018717 13.8602 -0.33411 0.0092446 36 0.055278 44.5831 -0.09700 0.0007850 37 0.025604 19.4713 0.64974 0.0345048 38 0.027045 20.6551 -0.30808 0.0078849 42 0.029542 22.7151 0.31873 0.0084404 43 0.036637 28.6275 0.30267 0.0076204 47 0.017167 12.6087 0.27599 0.0063199 62 0.012332 8.72782 -0.31476 0.0081823 65 0.024179 18.3038 0.49327 0.0200466 66 0.040116 31.5584 -0.72318 0.0429398 67 0.012496 8.85912 0.26313 0.0057367 80 0.049909 39.9232 0.06519 0.0003546 85 0.022290 16.7617 -0.25582 0.0054407 86 0.022381 16.8355 -0.29274 0.0071161 87 0.009944 6.82568 0.25909 0.0055530 106 0.044460 35.2476 -0.32462 0.0087675 130 0.021191 15.8667 0.66297 0.0357309 131 0.126369 111.682 -0.69386 0.0399993 132 0.052073 41.7947 -0.39653 0.0130718 146 0.010851 7.54744 0.26542 0.0058296 168 0.012287 8.69229 -0.27878 0.0064330 176 0.047155 37.5531 -0.37465 0.0116695 177 0.042470 33.5531 0.44089 0.0161283 217 0.021992 16.5186 -0.28031 0.0065269 234 0.046511 37.0009 -0.20338 0.0034476 235 0.076349 63.3938 -0.05276 0.0002323 256 0.042095 33.235 -0.36887 0.0113081 257 0.080274 66.9929 0.16676 0.0023194 272 0.011685 8.21165 0.28717 0.0068193 327 0.018359 13.5705 0.28821 0.0068913 Average leverage of single data point = 0.0153649. (continues on the next page) Excessive Leverage, DFITS, and Cook’s D values in italic.

484

Table K.3 Influential points for hypothesis one (continued)

Panel C: Model using daily data (continued) Mahalanobis Row Leverage Distance DFITS COOK’s D 335 0.05006 40.0543 0.25033 0.005221 393 0.04780 38.1076 -0.07673 0.000491 413 0.02409 18.2347 0.29243 0.007103 436 0.04637 36.8804 -0.06182 0.000318 442 0.07672 63.736 -0.03144 0.000082 443 0.07589 62.9748 -0.12603 0.001325 444 0.07727 64.2353 -0.13218 0.001457 445 0.07470 61.8941 0.30710 0.007857 456 0.02584 19.6693 0.36268 0.010905 494 0.04710 37.5109 0.02547 0.000054 495 0.07260 59.9862 0.06853 0.000391 516 0.07908 65.8974 -0.37068 0.011441 517 0.05076 40.6661 -0.10999 0.001009 518 0.07615 63.2157 0.20277 0.003428 521 0.02019 15.0597 -0.33030 0.009041 522 0.04716 37.5593 0.34871 0.010114 585 0.05139 41.2101 0.59025 0.028830 586 0.02283 17.2056 0.64398 0.033823 587 0.03972 31.2273 0.29031 0.007013 623 0.01013 6.97882 0.25717 0.005472 627 0.01617 11.8073 0.29641 0.007280 654 0.04780 38.1122 0.11357 0.001075 655 0.07164 59.1185 0.21141 0.003726 673 0.02308 17.4135 0.36830 0.011234 674 0.03633 28.3721 0.38758 0.012470 694 0.02489 18.8884 -0.31824 0.008407 703 0.07468 61.8811 0.15856 0.002097 758 0.03217 24.9009 -0.44833 0.016641 759 0.05442 43.8368 0.03935 0.000129 760 0.07908 65.896 0.02335 0.000045 761 0.03281 25.4355 -0.47291 0.018502 774 0.01660 12.1518 0.37057 0.011338 775 0.02317 17.4821 0.40128 0.013319 777 0.08290 69.4202 -0.02678 0.000059 778 0.05791 46.8929 -0.06689 0.000373 779 0.08346 69.9384 0.09498 0.000752 Average leverage = 0.0153649. Excessive Leverage, DFITS, and Cook’s D values in italic.

485

Table K.3 Influential points for hypothesis one (continued)

Panel D: Model using quarterly data Mahalanobis Row Leverage Distance DFITS COOK’s D 19 0.141933 8.44565 0.76054 0.11057273 20 0.137918 8.13625 -0.66221 0.08496698 31 0.045455 1.73158 -0.71167 0.08595706 60 0.407693 38.251 -1.43693 0.39820474 Average leverage of single data point = 0.0847458. Excessive Leverage, DFITS, and Cook’s D values in italic.

Table K.4 Influencial points for hypothesis two

Panel A: Model using monthly data Mahalanobis Row Leverage Distance DFITS COOK’s D 10 0.49152 30.928 1.12271 0.12453303 26 0.16657 5.6252 1.10826 0.10186089 34 0.44419 25.403 -1.50888 0.21199466 36 0.55808 40.703 2.5698 0.56483247 Average leverage of single data point = 0.285714. Excessive Leverage, DFITS, and Cook’s D values in italic.

Panel B: Model using quarterly data Mahalanobis Row Leverage Distance DFITS COOK’s D 23 0.150704 9.13169 -1.41316 0.2788794 31 0.268098 19.8966 -0.75469 0.0939430 32 0.156596 9.60052 -0.70661 0.0806471 33 0.225137 15.5786 0.90226 0.1312186 57 0.10271 5.54188 0.70562 0.0780503 Average leverage of single data point = 0.101695. Excessive Leverage, DFITS, and Cook’s D values in italic.

486

Appendix L Sensitivity Analyses

Fig. L.1 Box-and-Whisker plot of residuals with major outliers removed

-25 -15 -5 5 15 25 RESIDUALS

Model using daily data with 4 major outliers removed

Fig. L.2 Outlier plot with major outliers’ elimination

Outlier Plot with Sigma Limits Sample mean = -0.143534, std. deviation = 6.35444 34

24 4 3 14 2 1 4 0

RESIDUALS -6 -1 -2 -16 -3 -26 -4 0 200 400 600 800 Row number

Model using daily data with 4 major outliers removed (red X’s)

487

Table L.1 Additional regression model estimation and tests results

34 j MAtjtiitttt=+αβ∑∑ Per + δ Weekday, + λ SP500 + ζ FED + ϕ Hol + ji==11 (21)

++++++ϖ HS__ Extttttttφθψξε E BoM Event Event _ ED MA −1

Backward elimination regression, with all variables left in the model significant at the 0.05 level: T Standard Parameter Estimate P-Value Statistic Error Intercept 22.1014 0.0002 3.73514 5.91715 Per -0.04491 0.0000 -8.52012 0.005271 Per_2 0.00004 0.0000 7.21886 0.000006 Tue -3.04442 0.0002 -3.68739 0.825631 Wed -4.87279 0.0000 -5.97421 0.815637 Thu -5.24635 0.0000 -6.52683 0.803812 Fri -9.3084 0.0000 -11.5858 0.803433 E_BoM 12.6057 0.0000 15.0404 0.83812 SP500 0.01009 0.0104 2.56297 0.003938 MA_lag 0.10468 0.0005 3.45684 0.030282 ANOVA F value 86.57 0.0000 R2 (adjusted for d. f.) 51.032% N used (read) 740 (741) Durbin-Watson D 2.03524 0.3159 Chi-Squared (50 d. f.) 58.5726 0.189843 Shapiro-Wilk W 0.98280 0.120424 Skewness Z-score 3.26643 0.001089 Kurtosis Z-score 4.70102 0.000002 Kolmogorov-Smirnov 0.03136 0.469434 Modified K-S D 0.85703 ≥ 0.10 Cramer-Von Mises W2 0.16416 ≥ 0.10 Watson U2 0.12940 ≥ 0.10 Anderson-Darling A2 1.19931 ≥ 0.10 Kuiper V 0.05535 ≥ 0.10 Box-Pierce Test 24.2939 0.444899

488

Table L.2 Estimated autocorrelations for additional regression’ residuals

Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 -0.019282 0.036760 -0.072049 0.072049 2 0.057553 0.036774 -0.072076 0.072076 3 0.066627 0.036895 -0.072314 0.072314 4 0.061606 0.037058 -0.072632 0.072632 5 0.043876 0.037196 -0.072903 0.072903 6 -0.009975 0.037266 -0.073040 0.073040 7 0.031927 0.037269 -0.073047 0.073047 8 -0.003524 0.037306 -0.07312 0.07312 9 -0.031673 0.037307 -0.073120 0.073120 10 0.013436 0.037343 -0.073192 0.073192 11 0.017129 0.03735 -0.073204 0.073204 12 -0.043115 0.037360 -0.073225 0.073225 13 -0.042528 0.037427 -0.073357 0.073357 14 -0.04463 0.037493 -0.073485 0.073485 15 0.031350 0.037564 -0.073625 0.073625 16 -0.045995 0.037600 -0.073695 0.073695 17 0.028128 0.037676 -0.073843 0.073843 18 -0.027576 0.037704 -0.073899 0.073899 19 -0.002347 0.037731 -0.073952 0.073952 20 -0.032499 0.037731 -0.073953 0.073953 21 0.022204 0.037769 -0.074027 0.074027 22 0.005533 0.037787 -0.074062 0.074062 23 -0.066840 0.037788 -0.074064 0.074064 24 -0.003839 0.037947 -0.074376 0.074376

489

Table L.3 Estimated partial autocorrelations for additional regression’ residuals

Partial Standard Lower 95.0% Upper 95.0% Lag Autocorrelation Error Prob. Limit Prob. Limit 1 -0.019282 0.0367607 -0.0720498 0.0720498 2 0.057202 0.0367607 -0.0720498 0.0720498 3 0.069027 0.0367607 -0.0720498 0.0720498 4 0.061563 0.0367607 -0.0720498 0.0720498 5 0.039461 0.0367607 -0.0720498 0.0720498 6 -0.019576 0.0367607 -0.0720498 0.0720498 7 0.018427 0.0367607 -0.0720498 0.0720498 8 -0.010177 0.0367607 -0.0720498 0.0720498 9 -0.038463 0.0367607 -0.0720498 0.0720498 10 0.009095 0.0367607 -0.0720498 0.0720498 11 0.020773 0.0367607 -0.0720498 0.0720498 12 -0.040660 0.0367607 -0.0720498 0.0720498 13 -0.043352 0.0367607 -0.0720498 0.0720498 14 -0.045118 0.0367607 -0.0720498 0.0720498 15 0.035872 0.0367607 -0.0720498 0.0720498 16 -0.028456 0.0367607 -0.0720498 0.0720498 17 0.037468 0.0367607 -0.0720498 0.0720498 18 -0.020681 0.0367607 -0.0720498 0.0720498 19 -0.000313 0.0367607 -0.0720498 0.0720498 20 -0.031893 0.0367607 -0.0720498 0.0720498 21 0.022987 0.0367607 -0.0720498 0.0720498 22 0.005176 0.0367607 -0.0720498 0.0720498 23 -0.060646 0.0367607 -0.0720498 0.0720498 24 -0.006545 0.0367607 -0.0720498 0.0720498

Fig. L.3 Autocorrelations correlogram for additional regression model

Estimated Autocorrelations for RESIDUALS

0.6

0.2

-0.2 Autocorrelations -0.6

-1 0 5 10 15 20 25 lag

490